[DotGNU]4th/5th session summary

[S11001001]

Hello again, I missed session 4, but was there for session 5.

Brandon Bremen (absurd or absurdhero) has provided us with logs, as well as 
writing a nice Java(TM) program to get rid of the crap that the BX irc client 
put in it. This time, we have logs for the whole thing :-)

Anyway, we kick off with Mark Peters (skeeter) and ^CareBear discussing the UI 
of PhpGroupWare. Norbert Bollow (nb) points out that webservices are not 
necessarily free of charge. Some stuff about pnetlib non-standard classes... 
then there's talk about the development environment under DG. Hype surrounding 
Mono (announced self-hosting compiler Jan3, actual: Mar7). nymia announces port 
of treecc to Win32. David Sugar (dyfet) and Bill Lance (bill_lance) chat about 
connections between SEE, DEE, VRS, and auth.

^CareBear announces "if anyone have any Microsoft .NET questions, or something 
similar you can ask me :)" and leaves his email address with the crew. Then 
talk about the virtues of Unix DE vs. IDE. psiXaos pleads with them to provide 
list of decisions made during this meeting :)

Rhys makes comment of the evening on Mono..."I'm not sure it really solves the 
main problem - we use mono's code and they don't use ours.  It isn't enough for 
them to throw their code out and say 'feel free to contribute'.  We are 
throwing code out as well - why aren't they using it?" in reference to Pablo 
Molaro's offer of the Mono C# compiler (mcs) to help with pnetlib development.

fitzix offers a decision for psiXaos to munch on..."HINT: EVERYONE HERE -- 
START THE PROCESS OF CONTRIBUTING TO PNETLIB! :)" fraggle bemoans Mono's 
probable desire to focus on the hardware environments of its newly contributing 
companies. Talk of starting to contribute to pnetlib (HINT: see HACKING file!)

Rhys introduces Southern Storm Software, aka "Rhys, Inc". And then, as things 
are bound to do on #dotgnu, a bunch of chat about music appears, mixed with 
garbage collection architecture. nb changes nick to nb_very_tired.

There is some complaint about Linus comments about idealists, and chat about 
pnet releases. We start with Java, and eventually say that everyone should 
program with either a soldering iron or assembly. Long lull...

Nice chat about what the SEE is, with Barry Fitzgerald (fitzix) interrogating 
(HINT: compare to inetd). Some complaint about not sharing memory from me, and 
finally, Rhys forgets to check dependencies, so pnet compilation and hacking 
the build dominate the remainder of the session.

I will not be covering the next sessions, because it is 0213 a.m. here (2h13 
after session 5).

--
By golly, I'm beginning to think Linux really *is* the best thing since
sliced bread.
        -- Vance Petree, Virginia Power

This log painstakingly formatted by, AbsurdHero

Log Started Sat Mar 16 2155GMT
<S11001001> well, session 3 was fun, and even tho rhys is here, I have to go
<S11001001> c-ya
<31mù> S11001001 ddress@hidden;30m] has left #dotgnu 
[]
<^CareBear> Milosch, hm...so you won't build phpGW into a "server" style thing 
that anyone can build clients upon?
<31mù> valen ddress@hidden;30m] has left #dotgnu 
[]
<^CareBear> I really don't like the webbased GUI of phpGW and it would be 
pretty neat to be able to build richclients against the server stuff
<skeeter> ^CareBear: Just the opposite... the real power behind phpGW is the 
backend...
<^CareBear> skeeter? .. yes, that's what I mean? .. I would like to see a 
WebService interface against the whole system, so I could build rich client 
apps against it
<^CareBear> stuff like calendering just doesn't work that good when it's 
webbased, and the same with document-management
<^CareBear> http://www.dotgnu.org/web-services.html  <- I see here that SMS 
service is mentionened, how would you go by implementing SMS services when it's 
not something available for free? .. at least not here in Norway
<31mù> diotalevi [ddress@hidden;30m] has joined 
#dotgnu
<31mù> KoHaN ddress@hidden;30m] has left #dotgnu 
[]
<31mù> SignOff xenoneus: #dotgnu ()
<nb> ^CareBear: Web services aren't necessarily gratis.
<^CareBear> our inbox system supports Fax, e-mail, SMS, and more and the SMS 
and Fax part is something they "rent" as a service kinda, where they pay a 
specific amount of money to get a certain amount of SMS messages (which is 
provided trough one of our partners)
<^CareBear> gratis? .. that's norwegian word for free
<^CareBear> ah, hehe .. 
<^CareBear> yeh, I understand you now
<nb> We tend to use "free" for free as in freedom, or free software only..
<^CareBear> ok
<dyfet> free is a bit overloaded in english...too many meanings
<^CareBear> the SEE is the runtime for stuff like VRS?
<asshi_j_moon> 41 people heh
<asshi_j_moon> certainly more than normal
<nb> ^CareBear: SEE is something that executes stuff.
<31mù> asshi_j_moon is now known as fraggle
<dyfet> ^CareBear: SEE is a deamon that offers a plugin architecture for 
components
<nb> For example, a richclient on the user's local machine.
<^CareBear> nb, I asked a little earlier here today, is SEE some standard 
definition or is it an open system?
<^CareBear> but I don't see the difference between pnet & see .. 
<dyfet> ^CareBear: the difference is that SEE can use pnet, and then execute 
applications with it, but not the other way around...it can also potentially 
use other things like a java vm in the same way
<^CareBear> with the .NET Framework "SEE" is included within the framework, 
ain't it going to be included in the pnet framework?
<nb> pnet builds (among other things) a plugin for SEE that runs IL-based 
portable executables.
<fraggle> how far developed is see yet
<chiraz> see <=> a secure JVM and PNET <=> java bytecode??
<31mù> nymia [ddress@hidden;30m] has joined #dotgnu
<31mù> elvstone [ddress@hidden;30m] has joined 
#dotgnu
<^CareBear> I see that pnet is implementing some of the CAS
<rhysw> CAS?
<nb> When you want to add a new bytecode system to DotGNU, such as e.g. Parrot, 
then that can be done as another plugin for SEE.
<^CareBear> the reflection classes are awesome,but it doesn't look like it's 
implemented in pnet yet?
<rhysw> System.Reflection exists in pnet, but System.Reflection.Emit doesn't 
yet.
<^CareBear> rhysw, Code Access Security, which is what works as the "sandbox" 
for .NET assemblies, it can be tweaked by the developers and is not as static 
as the sandbox for Java
<chiraz> ahh... now I understand... so SEE = generic execution environment with 
plugins that allow different IL-based portable executables to run because it 
knows how..?
<fraggle> it would be cool if someone could get mcs to compile under pnet
<dyfet> chiraz: yes.  Plugins include vm's which know how to do the actual 
bytecode execution 
<nymia> hello
<dyfet> nymia: welcome
<fraggle> because the mono people have been making lots of noise about making 
their compiler self hosting
<rhysw> ^CareBear: righto - just wanted to confirm that my understanding of the 
acronym is the same as yours (MS has too many acronyms in .NET)
<fraggle> if pnet could compile mcs as well it would be clearly at the same 
level of development
<^CareBear> rhysw, in most cases when I talk about .NET, it's the .NET 
Framework and I tend to write it that way :)
<rhysw> fraggle: as I mentioned on the developers list a day or so ago, once we 
get pnetlib compiling (hopefully soon), then mcs will drop out the bottom as a 
freebie.
<rhysw> fraggle: but whether compiling mcs is anything other than a "cool 
demo", I still don't know.
<fraggle> you obviously need system.reflection.emit to be able to make it work
<fraggle> is anyone working on those classes yet
<^CareBear> there is no major differences between mono and pnet?
<nb> ^CareBear: Lots of major differences.
<^CareBear> I might be wrong, but what's the difference between SEE and CAS 
(which is beeing implementend in pnet)?
<fraggle> ^CareBear: they provide similar functionality but they are 
architecturally quite different
<nb> pnet isn't married to a single language like Mono seems to be.
<^CareBear> fraggle, yepp, that's what I thought
<rhysw> fraggle: no one is at present - some of emit is in place, but it needs 
to be hooked through to the runtime engine.
<^CareBear> nb, mono will support more than just C#
<^CareBear> I think there was someone working on a VB.NET implementation
<rhysw> if someone wants to take on emit, be my guest
<dyfet> ^CareBear: I would argue lots of differences, starting with the "P" in 
pnet :)
<fraggle> rhysw: i may investigate it 
<rhysw> fraggle: great
<fraggle> i've been thinking of rewriting my bf compiler in c# 
<nb> ^CareBear: Yes, but unless you're talking of something that I don't know 
about, that's not directly related to the Mono project.
<fraggle> rhysw: no promises just yet though
<fraggle> i need to find some documentation on reflection.emit
<rhysw> fraggle: the emit code in pnetlib shouldn't be that hard, since the C 
level of pnet already does most of the hard work - no need to implement all 
that PE/COFF ickyness at the C# level.
<31mù> shark [ddress@hidden;30m] has joined #dotgnu
<31mù> collord ddress@hidden:b80:366:1:280:c8ff:feb9:880e] 
has left #dotgnu []
<rhysw> fraggle: the only documentation for reflection.emit that I know of is 
Microsoft's .NET Framework SDK.  Emit isn't part of the ECMA standard.
<^CareBear> Portale.NET will support code created with the Microsoft.NET tools?
<fraggle> i have a copy of visual studio.net beta 2
<^CareBear> (at least basic stuff)
<fraggle> although its not currently installed
<fraggle> would that have the documentation
<SmileR> fraggle: I got a dvd with beta2, it has the docs
<fraggle> ok thanks
<^CareBear> I had the RTM at 13 feb :)
<rhysw> ^CareBear: pnet should run code built by the MS tools, as long as it 
uses libraries that are present in pnet.  Compiling C++ code with VS.Net 
probably won't work, because it depends upon MS libraries that we don't have 
and have no documentation for.
<rhysw> e.g. Microsoft.VisualC.dll
<^CareBear> rhysw, ok .. that's good
<fraggle> hrm
<shark> Hey all of the M$ developer tools are migrating to include .NET, right?
<^CareBear> if either pnet or mono won't support the most basic code from the 
MS tools it will have a hard time getting acceptance I figure .. I wounder if 
Rotot will be a full implementation of everything that's built into the current 
.NET Framework or if it will be looking similar to mono and pnet
<nb> The other icky is point is whether we succeed in running applications that 
try to use passport.com for auth, right?
<nb> ^CareBear: Definately.
<fraggle> surely that isnt part of the ecma standard though
<absurd> hmm. might want to make the dotGNU one compatable
<absurd> why did I type one?
<absurd> oh nm
<rhysw> ^CareBear: it's not that we "won't support" the MS tools.  It's that we 
can't because the MS library code is not free (in the FS sense), and there is 
insufficient documentation to build our own from scratch.
<absurd> add some sort of passport emulation?
<31mù> SignOff lstrcmpi_n: #dotgnu (Read error: 
110 (Connection timed out))
<^CareBear> support for Passport ain't essential .. we looked at passport but 
decided to go for an custom authentification system (which the .NET framework 
have GREAT support for)
<^CareBear> rhysw, good to hear!
<absurd> right. I know there is gonna be a completely different system but 
later on, passport "emulation" could be added, no?
<rhysw> nb: we need to identify what API's are used in .NET to call out to 
Passport, and then redirect those API's to our own auth facilities.  It should 
be transparent.
<absurd> some sort of bridge between passport reliant software and GNU
<absurd> thats what i was thinkin
<nymia> i tried compiling ccs on VC++ but it failed on the assembly part. Any 
thoughts?
<nb> rhysw: We can do that yes...
<dyfet> nb, rhysw: agreed
<rhysw> ccs?
<nb> but it still doesn't guarantee that we can run each .NET program...
<31mù> bill_lance [ddress@hidden;30m] has joined 
#dotgnu
<absurd> uhuh
<nymia> waitasec...
<shark> will people who use VisualStudio.NET be able to code there programs for 
dotGNU out of the box? If not, will they have to use GNU/Linux to code for 
dotGNU?
<nb> the executable might rely on some specific data that it gets from 
passport.com as part of the process of checking that the user is authorized to 
run the program.
<^CareBear> rhysw, I would like to see both our client & server running on 
something other than Windows in the future so it's good to hear compability is 
a goal =)
<nb> shark: Definately.
<absurd> well, with the program is byte code so it wouldnt need to be 
recompiled assuming it doesnt rely on anything eclusively MS
<rhysw> shark: as long as VS.Net users restrict themselves to standard ECMA 
facilities or DotGNU-supplied libraries, and don't use MS "embrace and extend" 
API's, there should be no problem.
<nb> If they want to code for DotGNU, they'll definately be able to do so.
<absurd> I know I will. But then again, I dont intend on learning C#
<absurd> what are the benefits of C# over java?
<nb> The problem is whether we can get to the point where _all_ .NET code works 
without exception.
<rhysw> shark: unfortunately, I know from bitter experience that many 
programmers think that if their system has it, all systems have it.  I expect 
many, many, many months of "but it works on my system" from people who aren't 
using ECMA facilities.
<fraggle> microsoft break their own standards
<fraggle> heh
<nymia> i think it was libgc. tried compiling it using VC++ and it failed.
<rhysw> absurd: there are some nice language features, like properties, value 
types, and delegates, in C# that Java should have had, but didn't
<absurd> thats a nice way around any restrictions on them requiring them to 
publish standards
<absurd> hmm delegates?
<absurd> what do those do?
<chiraz> just reading up further on dotGNU: MS uses SOAP as the basis for their 
RPC... What is the main RPC for dotGNU, Jabber?
<absurd> properties im kindof familiar with
<rhysw> nymia: were you trying to build pnet with VC++?
<fraggle> soap as well surely
<nb> chiraz: There's someone working on an abstraction lib...
;33mabsurd/#dotgnu gets out the XML book and reads about soap
<^CareBear> absurd, most important is the compability of WebServices, I don't 
know and haven't checked, but have MS been following things like the WSDL 
standard to the full lenght?..
<absurd> no clue
<rhysw> absurd: delegates are an abstraction of function pointers.  They allow 
you to package up a reference to a method and pass it around as an object.  
Very useful for event handling, observer patterns, etc.
<dyfet> nb: XMLRPC and SOAP are the most likely to be widely used
<shark> i know this is far fetched, but it would be cool if there was a 
"VisualStudio.GNU" or something - basically a GNU replacement for all the 
visual studio tools, which would be free and only support standards.  If we 
could provide that for free, we could get tons of people writing standards 
compliant code... ;)
<nb> with the goal that your code uses an API of the abstraction lib and then 
it is capable of talking all the standard RPC protols.
<rhysw> shark: are you volunteering to write all that code? :)
<fraggle> heh
<absurd> hehehe
<dyfet> shark: glad you volunteered for that...
<absurd> well, i have an idea
<chiraz> hehe... but looking at a dev-environment like that, there is no longer 
any need for automake and libtool, is there?
<absurd> jedit could easily be modified to be like that
<fraggle> you could probably write some emacs scripts to do dotgnu stuff i guess
<chiraz> anjuta maybe?
<^CareBear> shark, SharpDevelop is an opensource IDE which is looking pretty 
good, it's Windows based though :-|
<shark> rhysw: haha.  unfortunatly, i only know some c and java  - and i hate 
to program for/ use windows... ;) 
<absurd> i think it already supports multiple languaes
<bill_lance> What came of the talk of an ISE some weeks ago?
<bill_lance> IDE
<nymia> yes, i'm trying to build it using VC++. the asm part was written in gas.
<nb> dyfet: yes... and KISS is a good principle... why use anything 
complicated in situations where XML-RPC does what you want.
<shark> ^CareBear: cool, thanks . *goes to look into it*
<absurd> and can easily add a C# plugin for jedit
<rhysw> chiraz: automake/libtool/etc will never go away, because IDE's will 
never replace the need for automated command-line builds.
<nb> dyfet: we may want o add support for CORBA though...
<absurd> is there C# syntax support in emacs?
<dyfet> nb: yes we probably do
<nb> it has a significant installed base in corporations.
<^CareBear> http://www.icsharpcode.net/OpenSource/SD/default.asp  <- Sharp 
Develop
<fraggle> absurd: use java-mode :P
<dyfet> absurb: is there not?  I would be surprised if there is any language 
not supported in emacs :)
<absurd> ya i know. it probably is
<absurd> ill have to look
<rhysw> ^CareBear: I believe that the Mono guys are working with SharpDevelop 
to integrate mcs into it.  We'll get that for free once Mono gets it working.
<absurd> and btw, what file extention is used for c# files?
<chiraz> nb: strictly speaking, I suppose there should be ways to interpret 
both XML and IDL and join the two together somehow... but that's probably not 
the academic method :)
<rhysw> .cs
;33mabsurd/#dotgnu loads up xemacs
<absurd> there doesnt seem to be C# support!
<absurd> there is going to have to be when pnet and mono get a bit more well 
known
<rhysw> since I'm a vi person, someone else will need to contribute the emacs 
lisp ickyness to support C#
<nymia> also, just wondering who has the lead in development? is pnet ahead in 
development to mono?
<bill_lance> rhysw:  don't like lisp. eh?
<fraggle> heh what
<fraggle> dont like lisp? what kind of gnu developer are you?! :P
<rhysw> bill_lance: don't mind lisp - but don't like emacs - I want my editor 
to edit, not psycoanalyse me. :)
<bill_lance> rhysw:  It does sort of lead, doesn't it  :)
<nb> nymia:
<rhysw> nymia: comparing pnet and mono to judge who is "ahead" is apples and 
oranges.  Their C# library is more advanced, but our compiler design is more 
flexible.
<31mù> blazko [ddress@hidden;30m] has joined #dotgnu
<nb> nymia: It's hard to tell how far Mono really is.
<nymia> thx for the info, not really in know about that stuff.
<31mù> habbe [ddress@hidden;30m] has joined #dotgnu
<nb> In mid December they prodly claimed that they'd be self-hosting by the end 
of the year...
<bill_lance> rhysw: have you or any of your group looked at how PNet can be 
pluged in to SEE or VRS like servers yet?
<nb> Rhys and I questioned that... it certainly did't look possible at the time.
<absurd> im reading all about RPC and XML-RPC right now. Im glad i got this 
"XML for Java" book
<nb> then on Jan 3 they claimed that they had done it, and were ready to move 
development to Linux.
<31mù> SignOff erik_: #dotgnu ("Klienten 
avslutas")
<fraggle> well if theres one thing the mono guys are good at its generating hype
<nb> Now their website http://www.go-mono.com/ says: Mar 7, 2002 - Today Paolo 
got the MCS compiler compiling itself on Linux completely for the first time! 
The resulting image still contains some errors, but the
<fraggle> hype gets interest and interest gets developers
<nb>                   whole compiler process goes now
<skeeter> id you really say that? :)
<rhysw> bill_lance: I haven't looked at it yet.  It's waiting for a volunteer.  
I don't expect it will be hard.
<nymia> Hey Rhys, thanks for making treecc.
<bill_lance> rhysw: we'de like to plugin Apache and PNet as soon as we have a 
working clutster protoype
<rhysw> bill_lance: great - let me know if you need some help - I should be 
able to find a day or two to work on it, if someone can give me a spec of what 
they require.
<nymia> got it ported to win32 so I now have a single tool for two envs.
<rhysw> nymia: are you using treecc for a particular project, or just to build 
pnet?
<nymia> Linux & W32
<nymia> not work related.
<bill_lance> rhysw: Ir would be into a perl prototype first,  Pnet is C 
executables?
<nymia> not just pnet.
<rhysw> bill_lance: yes, pnet is all C, but it shouldn't be hard to glue it to 
perl (I'm no perl expert, but it does have a way to call out to external .so's 
right?)
<31mù> SignOff shark: #dotgnu ("time for 
homework! thanks for the info, guys.")
<bill_lance> rhysw: perl has several ways of hooking up to C
<nb> rhysw, bill_lance: yes, no problem to call C code from Perl, or vice versa.
<31mù> diotalevi ddress@hidden;30m] has left #dotgnu 
["Client Exiting"]
<^CareBear> but any phpGW guys here .. will you guys be building a feature 
complete WebService interface against the services provided by the phpGW 
software?
<nymia> think perl was written in C.
<bill_lance> The more I thinks about it, the more useful a VRS node might be 
just all alone, to consolidate all the services functions even for a single user
<bill_lance> At least that part of it that fulfills service requrests
<bill_lance> How about a servers that consolidated http and PNil seamlessly
<bill_lance> Pnet
<nymia> is it possible to rewrite the asm part in C?
<rhysw> nymia: which asm parts?
<nymia> can't exactly locate it, give me a minute.
<31mù> psiXaos [ddress@hidden;30m] has joined #dotGNU
<rhysw> bill_lance: definitely doable, although there would be overlap with 
SEE.  Perhaps a http control protocol for SEE?
<dyfet> rhysw: perhaps that can be done in reference to the token spec...hmm
<bill_lance> rhysw: well good, cause that's what we hope the Services Manger in 
VRS will be
<bill_lance> But is might be a useful stand alone for private servers as well.
<^CareBear> rhysw, what's the difference between the Code Access Security 
classes in the pnet framework and the SEE?
<31mù> thePooka [ddress@hidden;30m] has joined 
#dotgnu
<rhysw> ^CareBear: CAS is there for ECMA-compliance.  SEE is a set of policies 
for what CAS will and won't allow.  (In theory at least - still have to write 
the code :) )
<blazko> hi all, was invited today by t3rmin4t0r to join here. got a silly 
question...
<rhysw> ^CareBear: i.e. CAS is the API, and SEE is the implementation (more or 
less)
<blazko> can i start hacking a bit by using the actual 0.3 or 0.28 C# compiler 
right out of the box?
<^CareBear> hm.. k
<blazko> it is not about using dotGNU features as such but about doing with C# 
in first place
<blazko> ... for the first...
<rhysw> blazko: simple C# applications are currently possible.  A large chunk 
of the language is present, in varying states of bugginess.  Experimenting with 
"Hello World" like things is definitely doable.
<blazko> rhysw: thx, are the diffs between the 0.28 debs and the 0.3 tarballs 
that large thus it is quite okay using the pnet.deb from january?
<^CareBear> rhysw, so SEE include something similar to the "Microsoft .NET 
Framework Configuration" mmc control?
<blazko> rhysw: additionally, think the tarball can be installed to /opt or 
will there be problems?
<rhysw> blazko: there were a ton of compiler changes between 0.2.8 and 0.3.0, 
so the 0.2.8 debs may not be the best for experimentation.  The tarball should 
be installable anywhere using "./configure --prefix=/opt".
<blazko> rhysw: big thanks. although i am asking stupid things here. i will 
start right now and will be back after some experiences. although coming from 
the Perl/XML-XSLT/Object Pascal corner, perhaps i can help in the future when 
having gathered experience
<rhysw> blazko: no problem
<fitzix> hello Rhys :)
<rhysw> hi barry
<fitzix> Just caught up -- looks like some interesting discussion here
<blazko> last question: is this channel busy all the times or only at the 
weekend?
<bill_lance> dyfet:  I would like to explore how SEE might fit with both DEE 
and VRS
<fitzix> is there need for behavioural model of how SEE should work to be 
hacked out?
<dyfet> bill_lance: I would be happy to talk about this.  I know John is 
looking at how it fits into Freport.
<bill_lance> dyfet: it is essentiall a wrapper arounf pnet?
<dyfet> bill_lance: no, it is not essential at all to use pnet
<bill_lance> dyfet: could you eaplain it please?
<dyfet> bill_lance: it is essential for implimenting a security model around 
pnet and other bytecode engines we might use.
<bill_lance> yes
<bill_lance> how does SEE fill that need
<31mù> SignOff mastro: #dotgnu ("[x]chat")
<dyfet> bill_lance: SEE is more a part of the client side vision of how dotGNU 
will be deployed and provide for securing user applications.  That is SEE can 
do things like check and validate signed code, and control what applications 
are executed that might be brought onto the local client from a remote site
<dyfet> bill_lance: since it's a general purpose plugin framework, it can be 
used for other purposes as well
<bill_lance> rhysw: it is client side only?
<absurd> and it will control permissions to devices on the computer that 
programs have, right?
<elvstone> i just wan't to say that you guys that develop this shit rocks, keep 
up the good work, all of you, how long will this public irc meeting go on by 
the way?
<dyfet> absurd: yes, in regard to sandboxing
<rhysw> bill_lance: pnet is neither client nor server - it is an engine that 
can be used anywhere you want to embed it.
<absurd> ok. and when will support for java start/when did it start
<bill_lance> elvstone: two days, I think
<elvstone> bill_lance: ok.
<absurd> its been 12 hours and there are still another 24 hours to go on the 
meeting
<rhysw> elvstone: thanks for your support - the IRC meeting is continuing all 
weekend, with regular meet times every 4 hours (the next is in 2:45 from now)
<dyfet> rhysw: and pnet can be embedded on it;s own as a bytecode engine 
anywhere, so it can serve as an extension system for apps as well...
<chiraz> just a small q: in CORBA, there is wide support for 'naming' and 
'trading' to find services. How is service discovery supported for dotGNU 
eventually... Is it purely by RLS?
<rhysw> dyfet: yup
<dyfet> rhysw: as well as being highly portable :)
<rhysw> dyfet: yup again - portability is its defining characteristic
<absurd> thats why its called portable.net
<fitzix> chiraz: RLS is the optimal way, IMO.  An RLS string can hold almost 
anything that is needed for find services.  It's a valid URI on top of that, 
which the W3c likes
<dyfet> rhysw: has anyone talked about doing a mod_pnet for apache yet? :)
<absurd> shouldnt it be pgnu? portable.gnu?
<fitzix> thanks to contributions by sbp, AaronSW, and Dan Connely for that
<chiraz> fitzix: but what about when the user does not know yet where to find a 
service... i.e., something like Yellow Pages? 
<rhysw> absurd: the pnet name predates the merge with dotgnu
<fraggle> less quarreling over names, more code
<absurd> haha.
<absurd> but portable.net is gnu and its not .NET
<^CareBear> chiraz, UDDI? :)
<absurd> even though dotGNU came after it. whats wrong with you?!
<fitzix> chiraz: Well, that service (like a UDDI type of service) could be 
created and could pass them in the form of RLS strings, that would be the best 
solution, IMO
<absurd> ;)
<chiraz> fitzix: sorry, suppose a user knows a service existed somewhere in a 
domain, but does not know which one to use (based on the assumption one should 
be selected on a given preference)
<rhysw> dyfet: there may have been talk about it - I've thought about it 
myself.  But it comes back to the volunteer thing - I'm running out o hours in 
the day to implement every good idea.
<absurd> speaking of implementing ideas, i should be programming
<chiraz> UDDI, will read up on that, thanks
;33mabsurd/#dotgnu rotates 40 degrees towards his laptop
<absurd> hmm I need a super-efficient way to remove all markup tags from an 
html document
<rhysw> rm file.html
<absurd> is there any easy way?
<chiraz> lol
<absurd> LOL
<fitzix> chiraz: The two services on that domain would, ostensibly, have to be 
given separate remote IDs for publication.  However, if you can think of 
something that the RLS is lacking -- we can see if we can hash it out for an 
update of the spec.  Of course, if you had two of the same service on one 
system, that could theoretically be dealt with in the token segment
<absurd> it needs to take less than 1/4 second to remove the tags. 
<absurd> in java
<fitzix> hehe
<bill_lance> what's RLS?
<absurd> I have the webpage in String and char[] form. Now I need to convert it 
to plain text w/o markup
<fitzix> chiraz: Of course, the problem there is that the accepting server 
would have to parse the tokens pre-discovery... which could be problematic :)
<chiraz> fitzix: I think UDDI caters for it. Basically RLS = Naming in CORBA 
and UDDI seems like "trading" in CORBA. Trading is a service that allows you to 
select the most appropriate service for your request.
<psiXaos> absurd: if there is < start erasing until > then wait again for <.... 
argggghh
<fitzix> http://www.dotgnu.org/spec/rls.htm
<nb> absurd: in Perl it'd be s/<[^>]*>//s;
<absurd> psiXaos, thats what i was thinking. But should I use a StringBuffer, 
or Strings?
<nb> assuming that the HTML is correct in the first place.
<31mù> x-virge [ddress@hidden;30m] has joined #dotgnu
<chiraz> An example is when you need to buy tires for your car. If you want 
racing slicks, you would possibly look for other suppliers than if you want 
regular ones.
<psiXaos> absurd: a buffer I think... actually 2 buffers
<absurd> nb: thanks for the great idea. Thats a regular expression, right?
<absurd> I can use regexs in java. hmm
<nb> absurd: yes
<bill_lance> absurd: actually that's a headach.  But perl programmers can't 
tell the difference
<psiXaos> absurd: isnt perl will be slow compared to manual copying
<absurd> but I think it would be too slow. I need to write a specialized method
<absurd> damn. there is no way around it :(
<absurd> I guess ill start writing
<absurd> Just like I wrote a method to convert a String to an Int
<psiXaos> absurd: you can use smt like duffs device 
<fitzix> chiraz: This, I think, would require a full UDDI type implementation.  
It would not be within the scope of RLS itself
<absurd> what?
<31mù> SignOff SidNL: #dotgnu ()
<psiXaos> absurd: just grep duff's device on net..
<absurd> ok
<chiraz> fitzix: no problem, I am just familiarising myself with dotGNU, so 
just looking into details here and there... thanks for the support
<^CareBear> I'm leaving soon ..
<fitzix> chiraz: Anytime :) -- if you wish, feel free to continue the 
discussion on UDDI... it could be about time we started to hash out adding UDDI 
support.  Any ideas?
<31mù> artimus ddress@hidden;30m] has left #dotgnu 
[]
<chiraz> hehe... you're talking to someone who got into dotGNU today :)
<^CareBear> if anyone have any Microsoft .NET questions, or something similar 
you can ask me :)
<absurd> ugh. that duffs device looks like some crazy C thing
<fitzix> chiraz: Today is always the best day :)
<chiraz> no day like today I guess... Today is the start of the rest of my life 
possibly :) haha...
<absurd> ^CareBear is leaving wow. I thought you were a virtual person who 
lives inside the IRC server
<^CareBear> huhu
<psiXaos> absurd: It is a fast memcpy thing. But I think it can be modified for 
a copy-and-remove code.. just a thought :P
<absurd> ya i bet it could. but i would have to port it to java
<bill_lance> ^Carebear Would it be fair to say that as a MS experienced 
developer, a IDE is you biggest missing requirement in the .gnu approach?
<absurd> i cant even figure out how it logically works
<^CareBear> <- built many WebServices that is the same as those listed on the 
DotGNU website, and built custom authentification system built upon SOAP
<fitzix> chiraz: The start of the rest of my life was last June -- it always 
comes sooner or later :)
<nb> ^CareBear: Maybe you could leave us your email address?
<absurd> as soon as someone adds C# code to emacs, there will be an IDE ;)
<bill_lance> absurd: well, that's not what an MS experience person thinks of as 
an IDE
<^CareBear> bill_lance, I won't touch something unless it has a good and decent 
IDE, I'm a software architect for our company and it's I who choice what do use 
or not :)
<dyfet> bill_lance: you mean there is something other than emacs?  Oh yes, 
there is a "windowing" one..xemacs :)
<chiraz> fitzix: how do you envision how something like UDDI should be 
incorporated?  It sounds like some type of C# library to me (more some type of 
application library) that does lookups in Operator Sites?
<^CareBear> bill_lance, we need basic stuff like good source control that is 
highly integrated within the IDE, we need support for templates stuff and so 
forth, and we need an IDE that is easily extensible (we have some custom-built 
VS.NET Snap-Ins and we have built loads of our own Macros)
<fitzix> chiraz: Actually, I envision it as working sort of similarly to rpmfind
<psiXaos> xemacs is emacs with smt weird called X window(?).... pressing 
Ctrl+alt+backsp is enough to get only the emacs part :==)
<fitzix> the rpmfind model is actually a good comparisson in some ways
<chiraz> fitzix: so a separate tool that does the lookup and comes up with the 
RLS to use in the actual app?
<fitzix> to have a published list -- and then have a client to access that list 
in searchable format
<fitzix> chiraz: Yep -- that's sort of what I'm thinking
<fitzix> You could also have discovery mechanisms on the webservice server for 
publishing a list of services
<^CareBear> and one thing that would be very neat as VS.NET has, debuging of 
multiple languages right from the same tool at the same time, our current 
project has VB.NET (which will all soon be replaced by C#), C# and C++ code in 
our project, and I can debug and run everything right from the IDE, and I can 
even debug deployed code right from the IDE .. the debugging features of VS.NET 
is above everything else
<^CareBear> nb, address@hidden
<fitzix> then have those available to different authenticated users -- or public
<chiraz> but what about something, say... more related to DNS... something that 
returns the list in the app itself, which would allow for developers to 
determine how this list is presented to the user?
<nb> ^CareBear: thank you!!!
<absurd> that is one thing ive noticed. The debugging is integrated in VS
<chiraz> that way, you could make it into pushbuttons, links or whatever...
<31mù> SignOff kefka: #dotgnu (Read error: 104 
(Connection reset by peer))
<absurd> why not create a dselect and apt type system?
<rhysw> ^CareBear: yes, IDE's are great in the right hands.  However, it 
doesn't fit well with the Unix philosophy of "one tool, one purpose" (emacs 
notwithstanding).  The Unix world would build an IDE out of interlocking 
pieces, rather than as one monolith.
<31mù> kefka [ddress@hidden;30m] has joined #dotgnu
<absurd> well, thats all good and well as long as the bits and pieces work 
together seemlessly
<FrePort> emacs = one tool, one purpose (to make me smile everytime I use it) 
*giggles*
<absurd> lol
<bill_lance> rhysw: true ... but I might agree with ^carebear that is would be 
nice if someone actually started do just that
<fitzix> so, let's say I requested a UDDI.Get("dotgnu.fitzix.org") -- that 
would return a list of published webservices to the app
<^CareBear> I have never used emacs, but how does it compare against Visual 
SlickEdit?
<absurd> lol
<absurd> you dont want to know
<rhysw> bill_lance: yes, it would be nice.  Any volunteers? :)
<absurd> its the uberghetto console version of VS
<bill_lance> hahaha  ... the forever question
<fitzix> chiraz: Yep, that could be done -- but you'd need a UDDI component on 
the webservice server to handle requests and keep updates.
<chiraz> fitzix: something like that... and the app itself would be responsible 
for visualising the list to the user... as long as the list is made available 
in the right ofrmat...?
<plisken> slickedit is pretty nice.  i've got it for linux.  autocompletion and 
argument help is nice.  I always end up going back to vim though
<^CareBear> nb, I never got any answer, will there be a complete WebService 
interface implementation for phpGW? ..
<fitzix> chiraz: Yep -- that's what I'm thinking
<chiraz> fitzix: but isn't UDDI based on the assumption of having "Operator 
Sites"?
<absurd> what about jedit? its an opensource IDE
<chiraz> basically like services that contain registrations of business and 
services?
<nb> ^CareBear: I hope do...
<fitzix> chiraz: Yes, but no one said we couldn't deviate from that if it's 
optimal
<nb> but I'm not able to speak with any authority on this...
<absurd> and it has a plug=in structure which lets it follow the UNIX idealogy 
meantioned abouve
<fraggle> nvi!!~@"£
<nb> the person to ask about this is Dan Kuykendall.
;33mfraggle/#dotgnu starts a flamewar
<^CareBear> nb, phpGW has alot of the same stuff as we have as it's alot of 
basic features
<fitzix> chiraz: This doesn't keep that from happening -- it could even be an 
integral part of it
<chiraz> fitzix: but I'd say that Operator Sites are a useful thing to have... 
Because it allows you to register alike business with different uses that 
people can then find a la Yellow Pages, without the need for registration in 
the webservice specifically?
<fitzix> chiraz: Well, it's like google for webservices
<fitzix> chiraz: I'd agree
<chiraz> It's more or less directory services... hmmmm...
<fitzix> chiraz: think of it like this:
<^CareBear> nb, http://www.dotgnu.org/web-services.html  <- most of the stuff 
here is already something that's built in phpGW, but it's not currently exposed 
through WebServices
<absurd> ooo. phpGW with a webservices directory. that would be sweet!
<plisken> ok, i donwloaded the pnet compiler, the pnet libs, and treec. after i 
do my builds I should be able to compile simple c# apps?
<nb> ^CareBear: That's a page that I've been wanting to update for a looong 
time. :-)
<fitzix> User requests from opsite, opsite gets active lists from publishing 
webservice servers
<fitzix> or, the User could just harvest from the target server itself
<^CareBear> phpGW with WebService interface would be like a opensource 
implementation of .NET My Services which is just what we are doing ourself, 
which is both bad and positive .. right now we don't have an option, we have to 
do it ourself or wait
<fitzix> This allows for the whole thing to be scripted and then automated
<fitzix> great for distributed add-ons to DotGNU
<rhysw> plisken: you should be able to - see the "samples" directory in "pnet" 
for some examples.
<plisken> rhysw: cool
<chiraz> fitzix: how about pre-registration? I.e. register once and maintain 
that registration from that point onwards?
<rhysw> plisken: keep in mind that it isn't possible to build pnetlib with pnet 
right now (Real Soon Now, I promise), but there is a pre-compiled version in 
"samples" that you can use.
<absurd> are there any cross-platform IDEs for java that autocompletion of some 
sort?
<fitzix> chiraz: That would work as well
<chiraz> absurd: www.netcomputing.de (anyj)
<plisken> rhysw: great
<chiraz> absurd: linux one is free, windows one is commercial
<31mù> thePooka ddress@hidden;30m] has left #dotgnu 
[]
<31mù> carnow [ddress@hidden;30m] has joined #dotgnu
<chiraz> fitzix: definitely worth pondering about
<absurd> grr. the osX one isnt out yet
<chiraz> but it's a webservice that builds on dotGNU... before we can do that, 
it's required to have some sort of environment that can be used to test the 
XMLRPC calls against.
<plisken> absurd: slickedit
<fitzix> chiraz: Write it down and post it to the developers list -- see what 
people think :)
<chiraz> fitzix: np... will have a ponder...
<fraggle> i presume everyones seem brainfuck.net by now
<fitzix> chiraz: Agreed, but let's write this down -- it'd be a shame to lose 
it :)
<absurd> they all cost money :(
;33mabsurd/#dotgnu sighs
<fraggle> http://www.ecs.soton.ac.uk/~sdh300/stuffage/bf.net/
<absurd> I guess I'll have to write on myself
<chiraz> absurd: I'll have a look
<absurd> its on my list of software to write. A good java IDE
<absurd> that is free!
<psiXaos> so can someone summarize what has been agreed till upon on this 
meeting?
<fitzix> rhysw: Did you see my truce suggestion?
<fraggle> truce with who
<rhysw> fraggle: 404 not found
<fitzix> regrading the mono argument
<fraggle> hmm
<fraggle> works ok here
<rhysw> fitzix: yes I did.  But I'm not sure it really solves the main problem 
- we use mono's code and they don't use ours.  It isn't enough for them to 
throw their code out and say "feel free to contribute".  We are throwing code 
out as well - why aren't they using it?
<chiraz> absurd: AnyJ2.0 is still there, which is freeware. AnyJ3.0 is not 
freeware, but commercial
<31mù> SignOff carnow: #dotgnu ("changing 
servers")
<31mù> hobo [ddress@hidden;30m] has joined #dotgnu
<chiraz> I've used AnyJ1.4 and was quite happy. Some years ago they had a 
clause, because of the support of the Linux community, to provide it for free 
to Linux Users...
<31mù> bill_lance ddress@hidden;30m] has left #dotgnu 
["Client Exiting"]
<chiraz> Sad to see that they've taken it out for V3.0 though :(
<fraggle> heh
<rhysw> fraggle: got it now - IRC ate the "tilde s" in your URL prior to dh.
<fraggle> ah
<fitzix> rhysw: Yes, which is precisely why I think a new project and agreement 
needs to be made... I'm not sure it's a complete solution -- but it would give 
a little bit more of an official feel to the collaboration.  I think they'll 
say no though.
<31mù> SignOff hobo: #dotgnu (Client Quit)
<fraggle> well the mono class library is gpled isnt it
<31mù> carnow [ddress@hidden;30m] has joined #dotgnu
<fraggle> if the mono people want their library all bsd-licensed then they cant 
use it
<fraggle> but on the other hand the gnu people want to protect their freedom
<rhysw> fitzix: I agree that they'll probably say no, or say nothing at all.  I 
wish I knew what would break this deadlock.
<fitzix> fraggle: Yeah - it's a tough decision...
<fitzix> I'm more concerned about MS continually embracing an X11 licensed 
lib... but we may have no other choice
<chiraz> absurd: maybe this is something (Java OpenSource community) 
http://nisoft.orbitel.bg/freebuilder/download.html
<fitzix> that is, unless a whole bunch of people from here start contributing 
to pnetlib
<fitzix> HINT: EVERYONE HERE -- START THE PROCESS OF CONTRIBUTING TO PNETLIB! :)
<fitzix> This is essential
<fraggle> i think that ximian quite probably care more about support from intel 
and whoever the other companies that offered to help were than help from dotgnu
<31mù> SignOff nymia: #dotgnu (Read error: 110 
(Connection timed out))
<fitzix> fraggle: agreed
<fraggle> sadly
<31mù> SignOff carnow: #dotgnu (Client 
Quit)
<psiXaos> fitzix: what is now needed for contribution?
<FrePort> fraggle: i don't know if you were around when NB asked for a written 
promise from the Slippery Fish (aka Miguel) to commit to working with us?
<fitzix> psiXaos: Learn C#, download the existing pnetlib and pnet code and 
start hacking :)... 
<fitzix> heh
<fraggle> i dont think so
<fitzix> seriously, it will be much easier when pnet can bootstrap pnetlib -- 
Rhys, how far away are we from that?
<psiXaos> fitzix: hmm.. is C# a requirement?
<fitzix> psiXaos: The class libraries are written in C#
<psiXaos> fitzix: I have a natural irritation when smt comes from Redmond like 
C#.. 
<chiraz> I'm not sure about how to make an application SOAP or XMLRPC aware 
though, is that included or completely out of scope as of yet?
<rhysw> fitzix: hopefully only a few more weeks.  I've switched the pnet 
compile from "syntax check" to "semantic check" on my personal copy, and am 
banging through the bugs.  That should take about a week.  Then I will turn on 
"codegen" and bang through those bugs.  Then I need to go back and implement 
the missing expressions, statements, and declarations.
<fitzix> psiXaos: Me too, and I understand the feeling - but I like to think of 
it as just a language.  If we get enough velocity, we can take it from them! :)
<psiXaos> fitzix: hmmm... ok, I'll accept that for now and start to hack 
on..heads down :)
<absurd> hey, i finally have a method that deletes tags
<fitzix> rhysw: Great! :) - that's awesome, Rhys :)
<absurd> i have yet to test it, but oh well
<fitzix> psiXaos: That's awesome! You rule! :)
<31mù> carnow [ddress@hidden;30m] has joined #dotgnu
<fitzix> We need an army of coders to write this.. a damned code militia... 
teach your dog to code if need be!
<fraggle> ok but i need to get a dog first
<fitzix> fraggle: If you get one to learn -- breed it :)
<absurd> will cats do?
<31mù> SignOff dyfet: #dotgnu ("Client 
Exiting")
<fitzix> absurd: yes
<absurd> well, thats two more coders...
<fitzix> heh
<chiraz> what about monkeys?
<absurd> btw, what are we coding?
<absurd> monkeys are great. you can pay them bananas to code@
<fitzix> absurd: pnetlib, the Portable.Net C# upper level libraries
<absurd> oh. too bad i dont know C#
<fraggle> do you know java
<absurd> yup
<fraggle> close enough
<absurd> i heard they are similar
<fitzix> Seriously, though - :) Go out, buy a book on C# and get down to 
business :)
<rhysw> the problem with monkeys is that banging on typewriters does not result 
in good code in a reasonable timeframe
<absurd> but i like java :( not redmond's attempt at breaking java
<chiraz> what should pnetlib do?
<fraggle> heh 
<chiraz> is it the regular classlibrary that pertains to the system?
<sbp> hey, absurd could do it in Java, and fitzix can port it :-)
<absurd> hey!
<absurd> great idea!
<rhysw> chiraz: yes - it is the basic C# library needed for ECMA-compliance
<absurd> i can write the XML libraries
<31mù> spectra_cruz_alta [ddress@hidden;30m] has 
joined #dotgnu
<absurd> a SAX parser anyone?
<chiraz> rhysw: Where is this ECMA-compliance documented (so that I know what 
should be in there?)
<fraggle> thats probably quite easy i should think
<fraggle> just write hooks to libxml
<absurd> libxml is already done?
<fraggle> pnet can already interface to external c libraries
<fraggle> see pnetcurses
<absurd> nice. so what else is needed? just interface to all of the c libs!
<fitzix> http://www.programmingtutorials.com/  -- has a C# intro, I think
<absurd> ill consider it. I DO want to help dotGNU
<absurd> but right now I have a program to finish..
<spectra_cruz_alta> ALL: could you people set the current topic?
<rhysw> chiraz:    ftp://ftp.ecma.ch/ecma-st/Ecma-335-xml.zip
<rhysw> that's the class library specification
<fitzix> We all have to take care of our personal lives first -- I'm doing so 
myself... but I'm getting to a point where I should be able to start up writing 
code for pnetlib myself
<fraggle> hmm
<rhysw> run it through pnet's "csdoc2html" program to make it readable.
<chiraz> thanks!
<absurd> hey, I managed to write my HTML to Text method with only one 
StringBuffer
<absurd> it has a built in delete(int startIndex, int endIndex) method
<psiXaos> absurd: thats great! :)
<absurd> so i can delete the tags with one line once i find the brackets
<31mù> SignOff carnow: #dotgnu ("BitchX: now with 
38 percent fewer calories!")
<absurd> that method only took 17 lines. shouldnt take long to process the page
<absurd> I have a really really big problem with my program :(
<absurd> it can only process ASCII pages
<fraggle> http://www.tuneless.org/ecma/WD13Oct01.html
<31mù> SignOff spectra_cruz_alta: #dotgnu 
("Client Exiting")
<plisken> rhysw: does that .dll in the samples directory have network 
programming code?
<fraggle> thats the c# language spec although it isnt that bad for learning c#
<rhysw> plisken: nope - still working on that
<plisken> rhysw: ok
<absurd> anyone know how to detect a character encoding and decode the byte[] 
stream into a String?
<fraggle> infact i think you can buy a c# book on microsoft press which is 
basically part of the spec 
<plisken> can anybody recommend a good c# book
<fraggle> plisken : ^^^
<rhysw> the book is a printout of the spec.  Get the main copy here:
<rhysw>     http://www.ecma.ch/ecma1/STAND/ECMA-334.htm   (C#)
<plisken> ok
<fraggle> ah
<fraggle> ecma have a web version
<fitzix> plisken: I've been learning with the O'Reilly Programming C# book
<plisken> fitzix: yeah, think about picking up a book at borders
<fraggle> heh
<fraggle> the ecma website is hilariously bad
<fitzix> I once picked up a book at Borders -- then I put it back down :)
;33mfitzix/#dotgnu slaps himself in the back of the head for that one
<rhysw> fraggle: "hilariously bad"?  That's being charitable to the ECMA web 
site.
<fraggle> haha
<fitzix> heh - I'm still downloading those two pages, and I'm on DSL
<psiXaos> get the pdf one
<fraggle> <ecma_webmaster> i want a blue background. how will i do this? use an 
all-blue jpeg for the background!
<absurd> OReilly books are gauranteed to be good
<fitzix> ugh -- baby blue
<absurd> you cant screw up buying an OReilly book as far as I can tell
<fitzix> absurd: Just make sure to get one with code in it -- sometimes the "In 
a nutshell" series can be tough if it's a first book in a language
<fraggle> <ecma_webmaster> i want links that pop up when i hover over them. how 
will i do this? use a seperate java applet for each link to accomplish the task!
<absurd> 100% of the books ive bought in teh past few months are OReilly
<psiXaos> fraggle: ecma developers were too busy writing specs than choosing 
their webmasters!
<fitzix> brb
<rhysw> psiXaos: no, there are no ecma developers - just managers sitting in 
smoke-filled rooms deciding on technical standards that they know little about.
<fraggle> 
http://validator.w3.org/check?uri=http%3A%2F%2Fwww.ecma.ch%2F&charset=%28detect+automatically%29&doctype=%28detect+automatically%29
<fraggle> nice
<psiXaos> rhysw: heheh. this must be true
<blazko> can someone explain why samples/getenv.cs DOES compile? (means: where 
does it know of System.Collections)
<absurd> wow! my program has broken the 2000 line barrier!
<chiraz> rhysw: Are the class libraries themselves OS-specific (basically are 
these 1:1 with JRE's?)
<plisken> yeah, i've got a getenv.exe now. how do i run it?
<blazko> plisken: bin/ilrun
<plisken> blazko: ok
<rhysw> blazko: it gets it from mscorlib.dll in that directory
<absurd> may I now consider myself a "Real" programmer?
<fraggle> 
http://validator.w3.org/check?uri=http%3A%2F%2Fwww.ecma.ch%2F&charset=%28detect+automatically%29&doctype=HTML+4.01+Transitional
<fraggle> absurd: dont be absurd, oh wait
<blazko> rhysw: ah, logical, but thought it was for pure runtime, not compile 
time...
<FrePort> absurd: I dunno - do you have a big blue splotch on your forehead?
<rhysw> chiraz: pnetlib is OS-independent, but it relies upon native methods in 
pnet to some OS-dependent features
<absurd> lol
<absurd> whats with Southern Storm Software?
<absurd> who is that and what do they have to do with pnet?
<fraggle> um
<fraggle> heh
<rhysw> Southern Storm Software, Pty Ltd is "Rhys, Inc", basically
<absurd> ahh
<blazko> :-)
<absurd> figures
<rhysw> There are useful legal and tax reasons for having a company structure 
to work under
<absurd> hahaha. rhys is a tax evader!
<fraggle> limited liability?
<absurd> pretending to be self-employed?
<absurd> I found a really really cool class on SourceForge. It is a class that 
allows you to open up the default browser to a website on any platform
<rhysw> limited liability is the primary motivation.  And no, I am not a tax 
evader (judging by the tax office bill I just got after making $0.00 last year).
<absurd> cool.
<psiXaos> rhysw: what a storm :P
<chiraz> rhysw: are you fully dedicated to OSS development?
<rhysw> chiraz: yep
<absurd> wow. how do you survive?
<elvstone> what music do you guys listen to?
<chiraz> wow.. I've personally wondered about business models that would enable 
me to do this as well :).
<rhysw> absurd: interest on other investments
<psiXaos> elvstone: classical, jazz, death-metal. it is weird I know.. but this 
is offtopic
<chiraz> haven't really come up with any, other than starting a business that 
would solely focus on supporting OSS developments/installations in the industry
<elvstone> psiXaos: off topic is my lifestyle.
<rhysw> elvstone: if the radio plays it, I listen to it.  If the radio doesn't 
play it, I don't listen to it.
<elvstone> rhysw: thats awful! :)
<plisken> cool. just ran my first c# program
<absurd> if I have it in iTunes, i listen to it, if i dont, then i dont listen 
to it
<elvstone> but i guess it depends on what radio channel you listen to.
<absurd> I tend to listen to weird al yankovic and the blues brothers =]
<fitzix> elvstone: metal, rock, classic rock -- mostly metal... I'm big into 
NIN right now... sort of a resurgance of my teenage years
<elvstone> absurd: ok, thats also a pretty limited way of finding new music :)
<blazko> plisken: yepp, meetoo; time to start developing, eh?!?
<elvstone> fitzix: nice.
<absurd> well, I go on limewire every now and then
<absurd> and get new songs
<elvstone> ok.
<plisken> blazko: yep
<absurd> but that happens about every 6 months
<elvstone> now i'm done being off topic for the day, thanks.
<fitzix> hehe
<absurd> right now, I have "Superstition" in my head. I havent heard it for 
ever but, you know how that goes..
<31mù> SignOff x-virge: #dotgnu ("pop()")
<absurd> lets see if my source will compile =/
<elvstone> i just started listen to New Order, i'm _so_ into synth..
<absurd> i think im in for a big debugging session here
;33mabsurd/#dotgnu executes build.sh =/
<elvstone> for how long has C# been around?
<absurd> 7 errors
<elvstone> auch.
<psiXaos> elvstone: 2 months?
<elvstone> psiXaos: nah, it's been around longer than that, right?
<^CareBear> it was created by Anders, and I think it was one of the things that 
was in from the beginning 4 years ago
<psiXaos> elvstone: yeah..
<^CareBear> it was first available in a useable form when .NET hit beta 1
<rhysw> officially, it has been known for about 1.5 years.  Unoffically, it has 
been worked on for about 4 years or so.
<elvstone> ^CareBear: ok, so are there any serious flaws in the language that 
you guys know of?
<psiXaos> elvstone: just one: Redmond-design :P
<rhysw> it's object-oriented and garbage collected
<rhysw> :)
<elvstone> rhysw: i don't mind OO :)
<blazko> supposed i have installed pnet in /opt/dotGNU/, where to install 
mscorlib.dll? damned stupid from me...
<psiXaos> rhysw: yeah I want my garbage back
<absurd> garbage collecting is something i cant live without
<absurd> i wouldnt know what to do in C++
<blazko> absurd: try perl also :-)
<absurd> ugh
<elvstone> absurd: i'm not familiar with the term, since i'm not a developer.
<fitzix> lol psiXaos
<rhysw> blazko: /opt/lib/cscc/lib/mscorlib.dll
<fraggle> elvstone: garbage collection?
<elvstone> garbage collecting, sounds like a profession.
<fraggle> heh
<elvstone> fraggle: yea.
<fraggle> garbage collection basically means that like
<fraggle> you can create an object or allocate some memory
<^CareBear> elvstone, C# is great, it's designed by the same dude that did much 
work on Borland Delphi
<absurd> you dont have to destroy objects when you are done with them
<fraggle> and when you've finished using it you just forget about it and it 
magically disappears
<rhysw>  ... well that's the theory anyway
<elvstone> fraggle: ah, thats kind of nice, but aren't there a reason for c++ 
to not do this?
<fraggle> the magic bit is the hard part
<blazko> ^CareBear: you mean Object Pascal, not Delphi :-) ...
<absurd> in java, you have soooo many objects. I dont know what i would do. my 
code would be about double the size
<psiXaos> elvstone: garbage collecting is like keeping free memory-kids tidy...
<fraggle> elvstone: theres overhead in the magic part
<^CareBear> blazko, jau
<elvstone> fraggle: will there not be situations where you want to take care of 
destroying objects yourself?
<absurd> but you can manually call the garbage collector if you think it is 
inefficient at some point in your program to let objects sit around
<elvstone> absurd: thats neat.
<fraggle> elvstone: well
<blazko> ^CareBear: hehe, i am an ex-delphi-dev, now, windows is /dev/null on 
my sys
<fraggle> having the ability to do both causes even more problems
<^CareBear> elvstone, you can destroy objects yourself, but that doesn't mean 
they will be destroyed .. but you won't have access to them anymore
<elvstone> fraggle: i can imagine.
<absurd> hey, im down to 2 errors
<blazko> rhysw: thx, works like a charm now!!! i have not dreamt of to get 
C#/pnet stuff working that early
<absurd> and i dont know how to fix either of them =/
<rhysw> there are two kinds of "destroy": free the resources used by an object, 
and free the memory used by an object.  In non-GC languages like C++, the two 
can be done with the same construct.  In GC languages, you have to arrange for 
resource frees some other way.
<^CareBear> elvstone, obj = null will give a message to the GC that the object 
is available for deletion, but you can't be sure that the GC will destroy it 
right away
<absurd> can anyone help me? Does this statement look wrong:  
if(char.toString().equals("<")) { do something; }
<psiXaos> gargage collector is for untidy programmers.
<rhysw> char is a reserved keyword
<absurd> well, i renamed it for you. 
<absurd> the variable is actually called c
<psiXaos> something is a reserved keyword :)))
<^CareBear> absurd, should be Equals and not equals?..
<elvstone> i know the first 12 "newbie" hours are past, but is there some kind 
of installation instructions for installing and playing around in C# using 
dotgnu components?
<absurd> ^CareBear, ill try that
<31mù> bill_lance [ddress@hidden;30m] has joined 
#dotgnu
<^CareBear> launching VS.NET
<fraggle> make install?
<fraggle> heh
<absurd> the error says, "char cannot be dereferenced"
<absurd> it points to the c.toString() part
<chiraz> I'm trying to compile pnet, but lex is missing on my machine. Where do 
I get that or how do I compile alternatively?
<elvstone> fraggle: ok, hm, i found the place to start looking now..
<rhysw> I'm assuming that you have "char c" declared someone.  If so, then it 
isn't an object, and so you must use if(c == '<') instead.
<sbp> absurd: c is obviously a string with a single char - you can't call a 
method on it
<blazko> chiraz: on debian, use flex.deb, otherwise there's a flex.rpm
<blazko> chiraz: or tarball, of course
<fraggle> surely you can call methods if you box it
<rhysw> chiraz: you'll need versions of lex and yacc (e.g. flex and bison), 
otherwise building pnet is impossible
<plisken> blazko: just wrote and compiled my first c# prog:) now I should 
probably use binfmt so bash can call ilrun
<absurd> sbp, I need to convert it to a string then
;33msbp/#dotgnu usually feeds stray chars to a Character class instance
<rhysw> fraggle: only in C#, but I believe that absurd is writing his code in 
Java
<fraggle> ah ok
<31mù> bill_lance ddress@hidden;30m] has left #dotgnu 
["Client Exiting"]
<absurd> java it is!
<blazko> plisken: congrats, but tell me about binfmt; what does it do? kinda 
bash-wrapper?
<rhysw> plisken: when you figure out how to make binfmt work, let me know, as 
I've had trouble getting it to work.
<absurd> then, i need to make c a char object instead of a simple datatype
<plisken> rhysw: ok. i've actually never done it
<sbp> Character nc = new Character(c); 
<fraggle> isnt binfmt the thing which detects the type of files and lets you 
run them
<fraggle> like you can set up wine to run programs when you type ./foo.exe
<plisken> blazko: it should be able to look at the first few bytes of CIL and 
then invoke ilrun
;33msbp/#dotgnu reckons that there should be a better way, but isn't 
conversant enough in the mysteries of Java
<rhysw> binfmt is a linux kernel extension that allows you to say "if you see 
an executable with this magic number, process it with this here program".  
Basically, plisken wants to redirect IL binaries to ilrun automatically so he 
doesn't have to type "ilrun blah.exe".
<absurd> sbp, thats just what i was looking at in my book! lol
<absurd> thanks spb
<sbp> heh, cool
<sbp> np
<blazko> plisken: wow, that's cool. think that prooves superiority of GNU/BSD 
about windoze, hehe
<absurd> but can I apply a char value to a Character?
<blazko> plisken: what did you write, a C#-Hello World?
<fitzix> rhysw: I used to have that working -- say, 1 1/2 years ago... let me 
see if I can find something for you
<absurd> hmm. oh nm. i get what to do
<fitzix> only with wine
<fitzix> hmm..
<FrePort> blazko: it's neat, but historically speaking the magic numbner file 
ID was first used in AmigOS on the Desktop and PrimOS on the mini.
<fraggle> the magic number for java classfiles is funny
<fraggle> 0xcafebabe
<plisken> blazko: for (int i = 0; i < 100; i++) ConsoleWriteline(i); hehe
<31mù> PsyCHotiC [ddress@hidden;30m] has joined 
#dotgnu
<FrePort> Correction. Not AmigOA. MacOS.
;33mnb/#dotgnu yawns
<psiXaos> fraggle: 0xcafebabe is waiting for the 0xdeadbeef !
<blazko> FrePort: that's true, used it - um - sixteen years ago. good things 
often die
<31mù> lebox [ddress@hidden;30m] has joined #dotgnu
<blazko> plisken: am new to it, but isn't this System.write()??? dunno
<nb> I'm getting very tired and need to go soon.
<plisken> blazko: actually Console.WriteLine
<rhysw> C# == Console, Java == System.  Thank you Microsoft for renaming 
everything.
<fraggle> heh
<blazko> plisken: or this. think i should buy the O'Reilly book...
<chiraz> should treecc-0.0.6 compile without problems? I'm getting an error:
<chiraz> yacc: e - line 43 of "gram_cpp.yy", syntax error
<chiraz> %pure_parser
<chiraz> ^
<chiraz> make[2]: *** [gram_cpp.cc] Error 1
<chiraz> make[2]: Leaving directory `/home/chilan/treecc-0.0.6/examples
<fraggle> it would have been funny if microsoft hadnt renamed it
<plisken> blazko: I think i'm going to get a c# tomorrow
<nb> Any questions on SEE or Bizplan before I leave?
<fraggle> and had left the classes named java.*
<^CareBear> rhysw, thanks Microsoft for getting it right :)
<absurd> hahaha. that would have beengreat
<plisken> blazko: book that is
<blazko> plisken: which one?
<plisken> ok. off to eat dinner and check out some c# tutorials
<^CareBear> well .. Im leaving to, and I won't be back here in weeks .. maybe 
longer
<^CareBear> maybe never
<plisken> blazko: don't know.  I'll have to see what they have at Borders
;33mnb/#dotgnu yawns
<blazko> plisken: see you then
<fraggle> i wonder if i could adapt the reflection.emit classes from mono
<31mù> plisken is now known as plisken_away
<^CareBear> I'll watch out for DotGNU .)
<rhysw> chiraz: use the CVS version instead - it's a problem with byacc.  You 
can ignore the error, as it only affects the C++ example program, and not 
treecc itself.
<^CareBear> bye, bye
<fraggle> is it ok to use mono code for dotgnu?
<rhysw> bye CareBear
<plisken_away> rhysw: great work, by the way.  I'll see what I can contribute
<nb> ^CareBear: Thank you so much for coming over.
<chiraz> bya carebear
<fraggle> ^CareBear: bye
<nb> I'v really appreciated your visit.
<^CareBear> ;)
<blazko> fraggle: well, from a license point of view, it think, yes
<rhysw> plisken_away: thanks
<psiXaos> Also I am going. Thanks all for the meeting. It was an enlightening 
and smooth discussion. See you later. 
<absurd> goodbye. have fun
<absurd> woohoo. it compiles
<fraggle> blazko: i only ask because all gnu software is supposed to be 
copyright-assigned to the fsf
<31mù> SignOff lebox: #dotgnu (Client Quit)
<fraggle> and if i dont own it then i cant assign it
<31mù> Virosss [ddress@hidden;30m] has joined #dotgnu
<31mù> nb is now known as nb_very_tired
<^CareBear> .. need me quick? .. #IF on Quakenet, or #C# on Undernet
<Virosss> tuka ima li bulgari?
<31mù> SignOff ^CareBear: #dotgnu ("bb real later 
= ) .. need me quick? .. #IF on Quakenet, or #C# on Undernet")
<31mù> psiXaos ddress@hidden;30m] has left #dotGNU 
["for sleep...."]
<blazko> fraggle: ? you have the copyright but it is under GNU's "protection"
<blazko> fraggle: so where is the prob? think there is no legal one
<fraggle> yes but
<nb_very_tired> fraggle: That is not correct.
<fraggle> if i base the code on the reflection.emit code from mono then i dont 
own it all
<nb_very_tired> Not all GNU software is copyright-assigned to the FSF.
<blazko> fraggle: maybe cos it is now licensed inder the X11 licence
<nb_very_tired> For example pnet is Copyright Southern Storm.
<fraggle> yeah
<fraggle> in general most gnu software is though
<absurd> wow. i went to babelfish and im trying to figure out what Virosss said
<nb_very_tired> When you contribute code, you have the option of 
copyright-assigning either to Southern Storm or to FSF.
<blazko> all one writes is automatically copyrighted to the author, if it is 
not stolen...
<rhysw> my machine is going crazy - brb
<absurd> and i tried Russian-->english and here is what it said:
<absurd> the mineral fertilizer of ima whether of bulgari
<absurd> LOL
<fraggle> haha
<nb_very_tired> fraggle: Yes... that's what makes most sense for packages 
with many contributors.
<absurd> i still cant figure out what language that is
<absurd> what do bulgarians speak?
<blazko> bulgarish? dunno
<31mù> SignOff skeeter: #dotgnu ("Client 
Exiting")
<absurd> i dont see it on babelfish so i dunno
<fitzix> LOL
<fitzix> Virosss: what language do you speak?
<Virosss> fitzix: Bulgaria
<Virosss> fitzix: Bulgarian
<fitzix> cool :)
<Virosss> fitzix: Who?
<nb_very_tired> You're rigth about that.
<Virosss> fitzix: opsi What:))
<Virosss> fitzix: Do you speak Bulgarian?
<sbp> blazko: lol @ bulgarish
<nb_very_tired> Good night everyone!
<sbp> c'ya nb
<31mù> Psy_ [ddress@hidden;30m] has joined #dotgnu
<nb_very_tired> It was great meeting you all!!!
<fitzix> Virosss: Nope, unfortunately I don't :(
<fitzix> 'night Norbert :)
<chiraz> night nb
<blazko> sbp: well, am from germany, dunno the correct writing... :-)
<31mù> AaronSw [ddress@hidden;30m] has joined #dotgnu
<sbp> I thought it was an excellent interpretation
<nb_very_tired> blazko: Where in Germany?
<fitzix> hey AaronSw -- long time no see
<AaronSw> howdy
<absurd> blazko, do you have an AgendaV3?
<blazko> nb_very_tired: greetz; near muenster/nrw
<Virosss> blazko: ti znaesh li BG we?
<Virosss> a we tuka samo hora ot En li ima we?
<blazko> ah, so many questions... :-)
<blazko> absurd: whaddayamean?
<nb_very_tired> blazko: I'm near Zurich, and going to bed.  *wave*
<absurd> have you seem them? its a linux handheld in germany
<31mù> Psy_ ddress@hidden;30m] has left #dotgnu 
[]
<blazko> nb_very_tired: n8!
<31mù> SignOff nb_very_tired: #dotgnu ("catch 
some sleep")
<absurd> i swear, my SF project is broken. It always says 0% activitiy
<absurd> activity*
<sbp> Hmm... "n8" doesn't work in English... :-)
<absurd> no wonder i didnt get it!
<absurd> how do you say 8 in german?
<blazko> absurd: nope, haven't seen; but who knows, am at CeBIT on monday
<sbp> n-acht
<absurd> oooh
<absurd> figured
<elvstone> mkdir ~/code/c#, let go! :)
<blazko> absurd: 8: acht (cannot pronounce it this way...)
<absurd> lol sorry, i do java only
<blazko> sbp: yepp, that's it
<absurd> blazko, got it
<absurd> bbl
<31mù> cvasilak [ddress@hidden;30m] has joined 
#dotgnu
<31mù> SignOff fraggle: #dotgnu ("BitchX-1.0c17 
-- just do it.")
<sbp> c'ya
<chiraz> <= only minutes away from his first look at pnet and pnet-lib
<blazko> seeya
<31mù> rhysw_ [ddress@hidden;30m] has joined #dotgnu
<fitzix> http://www.ectaco.com/online/diction.php3?lang=2
<blazko> chiraz?
<chiraz> yup?
<rhysw_> I'm back - my machine went nuts and I had to reboot it, complete with 
fsck
<31mù> SignOff rhysw: #dotgnu (Read error: 104 
(Connection reset by peer))
<31mù> rhysw_ is now known as rhysw
<blazko> rhysw: better try ext3fs or reiser... :-)
<31mù> SignOff Virosss: #dotgnu ("Please see this 
quit:)))")
<elvstone> what is the naming convention of C# source files? .cs?
<blazko> elvstone: yepp
<elvstone> okay.
<elvstone> now i just have to find something out to put in them :)
<31mù> fraggle [ddress@hidden;30m] has joined #dotgnu
<rhysw> blazko: once woody goes stable, I will be blowing away this box and 
then I'll upgrade to ext3 or reiser
<blazko> rhysw: well, besides those about 12 bugs woddy is quite stable, and - 
from my point of view - sid is more stable than any other distro...
<blazko> rhysw: but switching to ext3 should not be a problem these times
<rhysw>  ... and I need to find the time to do it of course
<chiraz> cool... I think I got it installed
<rhysw> blazko: it is if you have RedHat 6.2 and no disk space left. :)
<fitzix> I'm seriously thinking of changing to the HURD kernel after the latest 
debacle
<blazko> rhysw: i am quite happy with reiser since about three years
<rhysw> which debacle was that?
<fraggle> eww
<fraggle> i tried hurd a while back
<blazko> fitzix: so soon?
<fraggle> its very unstable
<fitzix> The kernel team bitkeeper thing and Linus calling all people who have 
ideological beliefs stupid
<fitzix> It's much better now from what I hear
<31mù> shukko [ddress@hidden;30m] has joined #dotgnu
<fitzix> But, the kernel is on the way, I fear - to becoming less and less free 
... even if only development-wise
<blazko> fitzix: eh? explain please...
<blazko> mean HURD?
<fitzix> blazko: no - the linux kernel I mean
<fitzix> :)
<fitzix> here's a link:
<fitzix> hmm... kerneltrap is in the process of crashing netscape... that's 
great
<31mù> shukko ddress@hidden;30m] has left #dotgnu 
[]
<fitzix> see http://www.kerneltrap.org for more info.. I'll have a better link 
in a second
<blazko> fitzix: i am surpised; but who cares motif-netscape? or mean 
mozilla-netscape?
<fitzix> try to view it with mozilla if possible
<fitzix> motif-netscape
<fitzix> http://www.kerneltrap.org/node.php?id=71
<fitzix> it came back up, just very very slowly
;33mchillywilly/#dotgnu pets galeon
<blazko> ...reading...
<fitzix> The same week, news came out that HURD is close to becoming usable
<chillywilly> it's all a misunderatnding by RMS
<31mù> SignOff PsyCHotiC: #dotgnu ()
<chillywilly> he though the Hurd was separate form debian
<chillywilly> er, thought
<31mù> pete031502 [ddress@hidden;30m] has joined 
#dotgnu
<fitzix> Well, you can use it without using Debian
<elvstone> anyone have the C# Lanuage Spec in a free format? i can only find MS 
Word .doc on http://msdn.microsoft.com/net/ecma/.
<fitzix> HURD is a Debian project
<blazko> elvestone: somewhere on ecma.ch in HTML
<chillywilly> it is a debian port
<fitzix> chillywilly: explain - if you will :)
<blazko> fitzix: nopem HURD is a GNU project... right?
<blazko> chilliwilly is right
<fitzix> blazko: Yes, but it's within the scope of Debian development last I 
heard
<rhysw>     http://www.ecma.ch/ecma1/STAND/ECMA-334.htm   (C#)
<rhysw> has pdf and postscript versions - take your pick
<blazko> fitzix: as BSD also
<31mù> SignOff neuro__: #dotgnu (Read error: 104 
(Connection reset by peer))
<elvstone> rhysw: thanks.
<rhysw> The "HACKING" files for pnet and pnetlib contain all of the links to 
the specs
<fitzix> blazko: huh?
<elvstone> rhysw: ok.
<chiraz> <== just ran his first HelloWorld executable on SuSE Linux 7.3 without 
any problems whatsoever :)))
<chiraz> pnet rocks! 
<blazko> fitzix: there is a BSD port also on its way from the debian guys 
(Debian "FreeBSD")
<fitzix> blazko: Ahh... heh - I didn't know that
<blazko> fitzix: read somewhere on debianplanet.org these weeks
<rhysw> chiraz: congrats
<chiraz> cheers... Now I am starting to see the picture what you were all 
talking about :)
<fitzix> Heh
<fitzix> cool :)
<fitzix> bbl everyone -- have some things I have to take care of
<blazko> bye, fitzix
<sbp> c'ya fitzix
<fitzix> see you guys later - nice meeting the people that I've never met 
before :)
<blazko> :-)
<fitzix> And welcome aboard :)
<sbp> Pff, I get it. Not nice meeting all us regulars
<chiraz> cheers, fitzix...
<fitzix> sbp: Heh - nice to see you :) (I've already met you!) :) ... we're 
going to go out and play billiards when I get to the isles, right?
<sbp> heh, absolutely!
<blazko> oki
<fitzix> later all
<fitzix> see you soon
<sbp> c'ya
<31mù> fitzix ddress@hidden;30m] has left #dotgnu 
["Client Exiting"]
;33msbp/#dotgnu can *feel* a .GNU pool competition in the air
;33mFrePort/#dotgnu reminds sbp to spell out the Dot... got to protect the 
trademark, bud. :-)
<sbp> ugh. But I'm lazy!
<blazko> rhysw: oh, thx for the binfmt explanation up there, over-read it.
;33msbp/#dotgnu maps his . key to "Dot"
<blazko> not "dot"??? :)
<rhysw> blazko: you're welcome
<blazko> oh right, according to savannah it is DotGNU
<rhysw> "we are the Dot in .GNU"
<blazko> :-)
<fraggle> a .gnu tld would be cool
<sbp> heh!
<rhysw> brb - I need to go organise some lunch
<31mù> rhysw is now known as rhysw_brb
<fraggle> i need to organise some sleep
<blazko> meetoo, seeya all next time (after reading C# tutors...:0))
<31mù> pete031502 ddress@hidden;30m] has left #dotgnu 
[]
<elvstone> hm, where should i put mscorlib.dll? the README.mscorlib doesn't 
say, and i can't find information about it on 
http://www.southern-storm.com.au/portable_net.html#documentation either..
<FrePort> Fraggle: it was discussed early on in the mailing list, but basically 
one would need to attract enough implementors to the platform that a registrar 
would see enough monetary renumeration to risk the application fee to ICANN.
<fraggle> goddamn icann
<31mù> AaronSw ddress@hidden;30m] has left #dotgnu 
[]
<blazko> elvstone: oh, i got it from rhysw: for me it was ./lib/cscc/lib/
<blazko> elvstone: i installed everything to /opt
<FrePort> Hmmmm... grand total participants today 49. More than twice the 
predictions.
<31mù> skeeter [ddress@hidden;30m] has joined #dotgnu
<elvstone> blazko: ok, thanks.
<FrePort> Welcome back skeets, reboot go okay?
<blazko> okay, i am off for today (sleep), seeya these days!
<31mù> blazko ddress@hidden;30m] has left #dotgnu 
["Client Exiting"]
<skeeter> yeah.. much better
<31mù> SignOff chillywilly: #dotgnu ()
<31mù> rhysw_brb is now known as rhysw
<rhysw> I'm back, lunch in hand
<elvstone> damn australian, it's 3 am here, *gasp* :)
<rhysw> what? I can't help living in the best country on earth
<rhysw> :)
<elvstone> hehe, of course you can't.
<chiraz> elvstone... where you at, holland? Germany?
<elvstone> Sweden.
<chiraz> Ah, cool country :)... Not Laholm by any chance? (THAT would be 
coincidence)
<elvstone> chiraz: no, but i've been there.
<elvstone> i think i have at least.
<chiraz> ah, ok... Just knew someone from there...
<elvstone> okay, i'm from Stockholm, the capital.
<elvstone> bbl.
<31mù> SignOff cvasilak: #dotgnu ()
<rhysw> We've just entered the next 4 hour block - welcome to all lurkers.  
This is the weekend-long "DotGNU-a-thon", to discuss all things DotGNU, pick 
the developers brains, and find out how you an help.
<fraggle> rhys where is the image output code in pnet
<fraggle> image/ ?
<rhysw> yep
;33mFrePort/#dotgnu *giggles*
<rhysw> the API is practically identical to the regular API (il_program.h), 
with additional functions for flushing everything out to PE/COFF format 
(il_writer.h)
<FrePort> He's Rhysss "Mr Pnet" himself, the man who leaps tall piles of 
requirements in a single bound, writes code faster than fifteen other 
programmers on meth, and... (I'm at aloss)
<fraggle> destroys entire armies with his coding kung-foo?
<31mù> plisken_away is now known as plisken
<FrePort> that'll do.
<plisken> is there a problem writing something like this.  int i = 1; 
Console.WriteLine("Number {0}", i);  I get an error on this, but the tutorials 
i'm looking at it, says it's legal
<rhysw> ... disguised as a mild-mannered RSI sufferer :)
<FrePort> Wonders whether a Pnet logo should be a visual pun off the rough 
cognate "peanut"
<31mù> robble [ddress@hidden;30m] has joined #dotgnu
<rhysw> plisken: it's probably a bug in pnet - are you using the .tar.gz or the 
CVS version?  Someone reported a bug with that a few days ago, which should be 
fixed now (except for string formatting not working yet).
<FrePort> yeah, but looking at the PM is "truth and justice" really the 
Australian way?
<plisken> rhysw: tar right now
<plisken> rhysw: I guess i should cvs
<rhysw> I'll be putting out a new tar.gz release in a week or two, once I've 
squashed a few more compiler bugs.
<31mù> metric [ddress@hidden;30m] has joined #dotgnu
<plisken> rhysw: i'll go ahead and get cvs after dinner
<FrePort> Later all. I need dinner and sleep. See yaz.
<rhysw> bye FrePort
<31mù> FrePort ddress@hidden;30m] has left #dotgnu 
[]
<31mù> plisken is now known as plisken_away
<31mù> SignOff ceb: #dotgnu (Read error: 110 
(Connection timed out))
<31mù> S11001001 [ddress@hidden;30m] has joined 
#dotgnu
<31mù> leopardi [ddress@hidden;30m] has joined 
#dotgnu
<rhysw> hello
<S11001001> how was session 4?
<rhysw> pretty good - got quite a lot of questions about pnet and pnetlib, esp 
how to contribute.  Also questions on C#.
<S11001001> my favorite :)
<31mù> leopardi ddress@hidden;30m] has left #dotgnu 
[]
<S11001001> rhysw: what do we have summaries of?
;33mS11001001/#dotgnu looks for himself
<rhysw> dunno
;33mspringer/#dotgnu yawns
<springer> Hi gentlemen; just got back to the house... Off to check the 
summaries.
<S11001001> springer: yeah, pretty quiet now...you should have been here earlier
<springer> Unfortunately, weddings have intervened.  I caught this "mornings" 
session.  I guess it would have been around 1400 UTC.
<S11001001> springer: 1. hows wiki? 2. only summary for 1st session. using this 
lull as an opportunity, I think I'll hack together a summary for 2, maybe 3
<S11001001> actually, with springer here, now might be a good time to talk 
about the development environment
<springer> For 1, I've started hacking it together.  To tell the truth, I'm 
doing it in Java right now, anticipating the IL translator.  I'm building it on 
top of a bunch of standard protocols (XML:DB especially, possibly JSP) so that 
we can plug-n-play libraries.
<springer> Is there a licensing issue with using SOAP, BTW?  I've seen some 
threads floating around that I haven't quite gotten to the bottom of.
<31mù> nymia [ddress@hidden;30m] has joined #dotgnu
<S11001001> hello, nymia
<springer> I'm not sure I'm the guy you want to talk about development 
environments with.  Give me vi (preferably vim) and a compiler, and I'm a happy 
camper.  Bean, what's a bean?
<S11001001> springer: a bean is an annoying java component
<S11001001> part of JavaBeans, an annoying java component model
<rhysw> part of Java, an annoying Sun invention that took out all of the 
interesting language features
;33mspringer/#dotgnu laughs
;33mS11001001/#dotgnu thinks java is better than C#
<springer> Yeah, it falls into the "sounded good at the time" category.
;33mspringer/#dotgnu doesn't have enough C# experience to comment on Java 
vs. C#
<rhysw> there is no "better" programming language - if there was, it would have 
been invented in the 1960's and we'd all be using it by now.
<fraggle> right tool for the right job
<S11001001> sometimes, I think, screw all these new langs, go back to C/C++
;33mspringer/#dotgnu uses and enjoys COBOL in his day job
<fraggle> springer: you scare me
<S11001001> what TZ is sbp in?
<rhysw> wimps, the lot of you.  If you can't program with a soldering iron, you 
aren't programming.
;33mS11001001/#dotgnu is an assembly veteran, oddly enough
<S11001001> when some people ported GCC to the TI-89/92+, I got mad and started 
complaining about "real coding"
<springer> Bah!  Punch cards, string and wooden pegs.  If they're good enough 
for Babbage, they're good enough for me.  *LoL*
;33mrobble/#dotgnu is still to see someone use punchcards as an input for 
cdrecord
<springer> S0C9 - Really?  I just put my TI-92 into storage.
<S11001001> a68k rules, TIGCC sucks
<rhysw> programming with a soldering iron: 1. don an asbestos glove.  2. grip 
the soldering iron by the hot end.  3. pound the keys on the keyboard using the 
plastic end.
<springer> or get metal keys.
<springer> and a very large fan.
<S11001001> but the TIGCC people were kind enough to show me the GPL for the 
first time
<31mù> quincy-illinois [ddress@hidden;30m] has 
joined #dotgnu
<S11001001> rhysw: sounds like random-monkey-signalling-net
<S11001001> I discovered GNU with the question, "what's the G in TIGCC?"
<fraggle> tigcc?
<S11001001> fraggle: some people ported an old 68000 GCC to the TI-89/TI-92 
platform
<fraggle> ti calculator stuff
<fraggle> nice
<S11001001> they're already complaining about no C++ support, and I complained 
that the C programs shouldn't go in the 89/ASM category
<S11001001> back to my earlier question: what TZ is sbp in?
<S11001001> n/m...sbp's clock doesn't work
<S11001001> Start of #dotgnu buffer: Sat Mar 16 20:26:04 2002
<S11001001> hours later...End of #dotgnu buffer    Sat Mar 16 20:26:04 2002
<rhysw> he didn't have to wait for the discussion - he sent his Delorean ahead 
to collect the logs and then he was able to publish immediately.  Mystery 
solved. :)
<S11001001> I prefer Kia's new model over Delorean
<S11001001> you only have to go 62MPH
<31mù> robble ddress@hidden;30m] has left #dotgnu 
[]
<S11001001> rhysw: should I just write a combined log for the whole thing?
<rhysw> where's you sense of adventure?  Anything less than 20 mph over the 
speed limit is walking.
<rhysw> yep - go ahead with the log
<31mù> SignOff quincy-illinois: #dotgnu ("Client 
Exiting")
<31mù> x-virge [ddress@hidden;30m] has joined #dotgnu
<31mù> Begbie [ddress@hidden;30m] has joined #dotgnu
<kefka> Oilers - Capitals Game is beginning
<kefka> can be viewed on CBC
<31mù> SignOff Zeroth: #dotgnu (Remote closed the 
connection)
<31mù> SignOff Begbie: #dotgnu (Remote closed the 
connection)
<31mù> SignOff chiraz: #dotgnu ("Client 
Exiting")
<31mù> SignOff x-virge: #dotgnu (Read error: 104 
(Connection reset by peer))
<S11001001> also, a whole chatblock appears twice
<31mù> roybatty [ddress@hidden;30m] has joined 
#dotgnu
<31mù> SignOff plisken_away: #dotgnu (Read error: 
104 (Connection reset by peer))
<31mù> x-virge [ddress@hidden;30m] has joined #dotgnu
<rhysw> it's pretty quiet
;33mS11001001/#dotgnu is finishing up the summary
;33mrhysw/#dotgnu is fixing nested class bugs in cscc
<springer> rhysw: Yep... Sorry to leave you, but I'm nodding off... "Real" life 
demands my attention tomorrow bright and early...
<31mù> springer ddress@hidden;30m] has left #dotgnu 
[]
<rhysw> I'm going to take an hour or so off before the next session - unless 
anyone has an urgent question?
<rhysw> bye all
<31mù> SignOff rhysw: #dotgnu ("using sirc 
version 2.211+KSIRC/981227-pre0.9")
<31mù> tropick [ddress@hidden;30m] has joined #dotgnu
<S11001001> hello anyone still awake. I am finished with the session logs for 
2/3, and I am wondering if any1 has logs from 4
<S11001001> hello, tropick, you just missed rhys weatherley :(
<31mù> fairchic [ddress@hidden;30m] has joined 
#dotgnu
<S11001001> things are cooling down too...I think it will pick up again @0600GMT
<S11001001> but I do know a few things, and there are probably some other 
developers still listening
<S11001001> if you prefer, I just posted a summary of 1400-2200GMT, and 
earlier, t3ermin4t0r posted the 1000-1400 summary
<S11001001> 
http://subscribe.dotgnu.org/pipermail/website/2002-March/000303.html -- 1st 
session
<S11001001> 
http://subscribe.dotgnu.org/pipermail/website/2002-March/000304.html -- 
second/third session
<31mù> NetSplit: irc.openprojects.net split from 
zahn.openprojects.net [08:00pm]
<31mù> BitchX: Press ^F to see who left ^E to change to 
[irc.openprojects.
<31mù> Netjoined: irc.openprojects.net zahn.openprojects.net
<31mù> x-virge [ddress@hidden;30m] has joined #dotgnu
<31mù> simc [ddress@hidden;30m] has joined #dotgnu
<S11001001> New Now Know How
<absurd> S11001001, I have logs from around 11 Pacific time
<31mù> cerb [ddress@hidden;30m] has joined #dotgnu
<S11001001> absurd: I was looking for logs from 6-10 your time <grins 
sheepishly>
<S11001001> oh, n/m
<S11001001> absurd: actually, I totally forgot what I asked. I need logs from 
2-6 your time
<31mù> SignOff cerb: #dotgnu (Client Quit)
<31mù> xenoneus [ddress@hidden;30m] has joined 
#dotgnu
<31mù> zag [ddress@hidden;30m] has joined #dotgnu
<31mù> fitzix [ddress@hidden;30m] has joined #dotgnu
<fitzix> hello
<S11001001> hey fitzix
<fitzix> hey there S11001001
<S11001001> it's been pretty quiet, and Rhys says he will return at 0600GMT
<S11001001> I wrote summary for session 2/3, and am waiting for the logs for 4 
from absurd
<fitzix> great -- what'd I miss?
<31mù> rhysw [ddress@hidden;30m] has joined #dotgnu
<fitzix> hey there Rhys
<rhysw> hi
<fitzix> what misc_binfmt issues were you having?
<rhysw> I could seem to get the darn module to load and then accept requests 
under /proc/wherever
<rhysw> couldn't rather
<fitzix> Hmm - what kernel?  What was happening?
<rhysw> error when trying to create the file - it was on RedHat 6.2 (2.2 kernel 
I spose) - I looked at the kernel source and there didn't appear to be any 
reason for it.
<fitzix> Red Hat default kernel?
<rhysw> yup
<fitzix> heh - Barry's rule #1 -- never use a Red Hat default kernel :)
<fitzix> That could be it... their default kernels were always a little funky
<fitzix> There was always something a bit broken
<S11001001> hey rhysw, couldn't stay away?
<rhysw> perhaps - I don't have the resources to retry just yet.
<fitzix> however, did you echo some rules to /proc/sys/fs/binfmt_misc ?
<rhysw> S11001001: my nap didn't take as long as I thought :)
<rhysw> fitzix: yup - that's when the open failed - root/nobody/me made no 
difference
<rhysw> the binfmt module was loaded at the time too
<fitzix> hmm...
<fitzix> [ -d /proc/sys/fs ] ?
<fitzix> err..
<fitzix> that was the dir for a 2.4 kernel -- I don't know if it used to be 
something different
<S11001001> fitzix: you missed a discussion about real programming
<S11001001> fitzix: I said assembly, rhys said soldering iron
<fitzix> S11001001: As opposed to fake programming, non-existant programming, 
novice programming - or all of the above?
<fitzix> ahhh
<fitzix> heh
<fitzix> Neither :)
<S11001001> <S11001001> they're already complaining about no C++ support, and I 
complained that the C programs shouldn't go in the 89/ASM category
<fitzix> real programming is converting everything to binary by hand -- then 
doing calculations manually modifying the memory address values
<fitzix> :)
<S11001001> those C-writing bastards were messing up our beautiful, elite ASM 
world
<absurd> lol ASM is useless in todays world! ;)
<absurd> assembly is for the nerds at intel :p
<fitzix> Unless you're writing virii, drivers, or code to run a plane engine 
--- you're right :)
<rhysw> tell that to fraggle with his brainfuck compiler - asm that even asm 
programmers can't understand :)
<fitzix> heheheh
<fitzix> I took a look at it a while back -- it's pretty funky... all pointer 
arithmetic.. the whole damned thing
<roybatty> how do i build treecc from cvs. there's no configure
<rhysw> run "automake", then "autoconf".  That will give you the config files.
<roybatty> thanks
<absurd> i always wondered how to do that...
<31mù> habbe is now known as dragonslayer
<31mù> dragonslayer is now known as habbe
<fitzix> It's cleaner that way... not pre-generating the configure script...
<rhysw> fitzix: I'd like to try out his brainfuck.net compiler, but there are 
no examples.  Although, that's probably the whole point with bf - writing 
programs is impossible.
<S11001001> yeah, CVS projects usually don't include results of autotools
<roybatty> it must take forever to figure out this autoconf stuff
<fitzix> rhysw: Heh - I have yet to figure out what it's actually supposed to 
do...
<rhysw> forever, a day, and then a couple of millenia after that - it's pretty 
weird
<absurd> mine does! i made a nice build.sh!
<absurd> :p
<S11001001> fitzix: didn't you send me the link to that book on the autotools?
<absurd> then again, java doesnt require ./configuring
<fitzix> S11001001: I don
<fitzix> t think that that was me
<rhysw> bf is one of those languages that is impossible to program in, like 
INTERCAL.
<absurd> what ever happened to rebol?
<fitzix> heheh
<fitzix> rhysw: is it intentionally made that way, or is there some other 
real-world purpose to it (judging by the name, I'd say no) ?
<rhysw> it's intentional
<absurd> ugh. too many windows open. over 10 of em =/  Good thing im in linux
<fitzix> hehe - what a brainfuck
<rhysw> precisely
<31mù> SignOff simc: #dotgnu ("ChatZilla 0.8.6 
[Mozilla rv:0.9.9/20020311]")
;33mS11001001/#dotgnu 's updatedb is running :)
<fitzix> If you've got a slow disk and a lot of files, that'll really chew up 
your system
<fitzix> hmm...
<fitzix> going back to binfmt
<fitzix> rhysw: what file did you try to write to?
<rhysw> fitzix: a few months ago when I tried binfmt, I was in the process of 
adding a "ilrun --register" option to make it easy to register ilrun with 
binfmt.  Because it didn't work, I took it out.  But I could put it back and 
you could debug it.
<31mù> SignOff elvstone: #dotgnu ("Hey!  Where'd 
my controlling terminal go?")
<fitzix> sounds good -- what's the basic premise?
<rhysw>  /proc/sys/fs/binfmt_misc/register  and  
/proc/sys/fs/binfmt_misc/DOSWin  were the files I was working with
<fitzix> cool -- looks good
<rhysw> The premise is that you would run "ilrun --register" in your system 
setup scripts or somewhere, and it would take care of the rest.
<fitzix> basically setup the files and put in the correct input.. pretty simple
<fitzix> OK - I'm game
<rhysw> Give me about 20-30 mins and I'll put it back in.
<absurd> ugh updatedb takes forever
<fitzix> Thanks Rhys :)
<absurd> when is the next session?
<fitzix> take your time 
<fitzix> Role call -- Who's here?
<S11001001> hallo
<S11001001> my role is doing Uri classes
<rhysw> I'm here, and my role is FFOD
<fitzix> FFOD? :)
<fitzix> Flying Frisbee of Death? :)
<S11001001> that's all I do, tho
<rhysw> Flying Fingers of Death
<fitzix> ahhh - cool ... even better than a frisbee 
<fitzix> but, few things are :)
<S11001001> when the VRS, SEE, DEE, Java plugin, cscc, and ilrun, and the DGDE 
are done, then Uri classes will probably be finished soon after
<fitzix> S11001001: Tis always better to do something than nothing
<fitzix> Question: why plugins for SEE?
<fitzix> Why not just exec directly
<S11001001> well, I also <ahem> maintain the tasklist
<fitzix> Can we find a real reason to have binary glue code?
<fitzix> S11001001: That's not even a remotely small job :)
<S11001001> fitzix: the SEE provides a thing that the outside world can talk to
<absurd> so YOUR responsible for maintaining the task list!
<S11001001> absurd: yep...what do you think! BTW, there are some changes I have 
that aren't up there yet, and I will post another one Real Soon Now
<fitzix> S11001001: I know what it does... but, why have plugins that implement 
an interpreter/compiler?  Why not just call the compiler?
<absurd> I think the list is very nice
<fitzix> err ... interpreter
<S11001001> fitzix: call it for what?
<absurd> and BTW, has anyone here every written a program that uses a thread 
pool?
<fitzix> S11001001: Let's say I have a IL program... I execute that by saying 
'ilrun <prog>' ... why implement a plugin that contains ilrun?
<S11001001> absurd: I got sucked in because I said DG should have a tasklist, 
and nb said "so are you proposing a tasklist project, and offering to be 
maintainer for it?"
<absurd> haha
<absurd> hmm. ill remember that next time i have a suggestion
<absurd> or next time I want to infultrate a developement project
<fitzix> What do you guys think of the new website design?
<fitzix> http://www.dotgnu.org/new btw...
<S11001001> fitzix: 'plugins'...they aren't really plugins in the netscape 
sense. It's kind of like X...the plugin and the SEE act as server and client
<S11001001> to eachother
<S11001001> the plugin provides the actual execution, and the SEE provides the 
data to execute, permission information, ....
<rhysw> fitzix: I'ved checked in the binfmt registration code into the CVS 
version.  See "engine/register.c" for instructions.
<S11001001> fitzix: I still hate that "GnU" pic
<31mù> t3rmin4t0r [ddress@hidden;30m] has joined 
#dotgnu
<S11001001> hey t3rmin4t0r
<t3rmin4t0r> hi S11001001
<fitzix> S11001001: Yes, I recognize that... I think that Rhys, didn't we have 
this discussion a few months ago? ... it passes info back to SEE ... Therefore, 
wouldn't it be better to implement most of the info that SEE would have passed 
back to it in libraries?
<rhysw> hi gopal
<fitzix> rhysw: cool :) -- I'll check it out now
<fitzix> hey gopal
<t3rmin4t0r> rhysw: hi
<S11001001> for SEE to work, it must be persistent...hence 'daemon'
<S11001001> What we'd love to have is a simple daemon which can honor local
<S11001001> requests and remote requests to execute bytecode which either
<S11001001> comes from the local disk or from a remote @acronym{SEE}
<rhysw> fitzix: that seemed to be the consensus at the time.
<S11001001> "
<S11001001> think of it as like inetd
<S11001001> SEE, you don't really need all the crap those servers bring with 
them all the time
<S11001001> but at the same time you need an app running that listens for 
connections
<S11001001> now most of the info, yes, libraries, but where do the libraries 
get the info? from the persistent SEE
<fitzix> Or, from the operation of the program
<S11001001> for example, ident data...how does that come from the operation of 
the program?
<fitzix> build the security extensions into the libraries and kill 2 birds with 
1 stone
<fitzix> S11001001: Wait a minute here -- I'm not saying that SEE has no 
purpose :) ... not at all...
<fitzix> S11001001: I'm just saying it's a bit overly complex for it's job...
<absurd> about the new site design, it looks great
<t3rmin4t0r> absurd: ???
<S11001001> fitzix: well, using IPC would make it complicated. However, I have 
some notes about ident, which may be invalid because I don't follow the auth 
list very well, but here we go
<absurd> the new dotGNU site design
<t3rmin4t0r> S11001001: I guess anyone writing a VM would be capable of doing 
IPC work
<t3rmin4t0r> absurd: dotgnu.org/new ?
<absurd> ya
<S11001001> well, like I said, the SEE should only be as complicated as inetd, 
except for the auth, which is provided by libs to both SEE and plugins
<absurd> I have one critique for the site:
<t3rmin4t0r> absurd: black ?
<absurd> Those cute little tab graphics at the top need to have predefined 
width and height so that the browser doesnt get all goofed up and ends up 
repositioning all of the graphics after it loads
<S11001001> "The idea of the daemon is that it keeps the identity persistent, 
effectively invoking thoughts of single sign-on, so this does not have to be 
rechecked. So if you invoke the API in your program, and there is already a 
running SEE, it can use that identity. Also, it can get separate identities, 
for use by separate programs. The main idea here is that you won't get confused 
about the ident, because there is a single daemon at the center.
<t3rmin4t0r> S11001001: the problem with SEE is that it has to be *designed*
<absurd> another great reason to have the SEE always there, is that the virtual 
machines dont have to reload everytime they are needed ala jvm
<S11001001> absurd: actually, they do
<t3rmin4t0r> absurd: they have to reload everytime,
<absurd> :( why?
<S11001001> absurd: I think the main problem with the jvm is that it uses eager 
class loading
<rhysw> yes, they do - each app instance has a different copy of the state
<absurd> no no no. sorry. i meant that the jvms DO have to reload
<t3rmin4t0r> absurd: to save memory
<absurd> oh. each app has to have another instance of everything?
<S11001001> absurd: according to "Java Performance Tuning", just to run a Hello 
World, the jvm loads something like 4,000 classes
<t3rmin4t0r> absurd: yes, except perhaps the SEE sessions ;)
<absurd> ya, Swing takes up about 20 megs in memory
<S11001001> t3rmin4t0r: I am making some progress towards design
<t3rmin4t0r> S11001001: that's becuase JAR files are loaded fully unlike 
directory organized files
<absurd> eww. so you cant cut down on class loading by specifying 
java.math.sqrt; instead of java.math.* ?
<S11001001> absurd: no swing involved here
<t3rmin4t0r> My program ideology is more for an evolution of a program
<S11001001> the JVM 1.3.1_01 on my system uses ~20MB min, no matter what
<absurd> i was using swing as an example of mega big memory usage by APIs
<t3rmin4t0r> absurd: no, since they are in the smae JAR file
<absurd> t3rmin4t0r, gotcha
<S11001001> absurd: no, the import statement only affects where the compiler 
looks for stuff
<t3rmin4t0r> absurd: but that can ease your compile times
<absurd> oh. ok. I thought there had to be some purpose for NOT always doing a 
.*
<31mù> SignOff x-virge: #dotgnu ("pop()")
<S11001001> absurd: if you ever look at a raw .class file, you will see that it 
puts in the whole name...where you have Exception, it will put 
java.lang.Exception
<absurd> ya, i should have thought about that
<t3rmin4t0r> absurd: or try disassembling it with Rhys's ildasm
<S11001001> rhysw: I was wondering about something dealing with ilrun loading
<S11001001> rhysw: 1. lazy class loading?
<S11001001> rhysw: 2. how about a .so system?
<absurd> but arent .so binary and therefore platform dependant?
<S11001001> one of the problems with Java (and most non-native stuff) is that 
the libs can't share memory like .sos
<S11001001> absurd: I just mean a parallel system
<absurd> oh
<t3rmin4t0r> absurd: We use a DLLImportMap Attribute to support .so & .dll 
provided they are the same libs
<absurd> i dont knwo much about that stuff :(
<t3rmin4t0r> eg glibc.dll on cygwin & glibc.so on GNU
<rhysw> 1. the loader brings in all of metadata at once, but performs class 
layout lazily.  I can change this, but haven't gotten around to it yet.
<S11001001> this is also in jpt; multiple JVMs can eat up memory because they 
don't share much
<absurd> well, they should learn not to be so selfish!
<rhysw> 2. it would be possible to turn ilrun into a .so - just haven't 
bothered because of the ickiness involved in libtool
<S11001001> what I am really talking about is support for memory sharing with 
portable code
<S11001001> I'm not so worried about the runtime itself, more with the library
<rhysw> ?
<rhysw> Do you mean pnetlib?
<S11001001> rhysw: yes
;33mabsurd/#dotgnu gets back to coding a method to count spaces in a string 
*ugh*
<absurd> talk about dull
<S11001001> although there are many apps running on my system, the glibc 
(2.2.5) they are linked to does not duplicate for each in RAM, right?
<t3rmin4t0r> S11001001: not for static methods
<31mù> carnow [ddress@hidden;30m] has joined #dotgnu
<S11001001> t3rmin4t0r: static methods pertaining to what?
<S11001001> hey Charlie carnow
<rhysw> it comes back to the same problem as jvm - there are certain things 
about classes that must be computed at runtime, and so they end up in the heap. 
 However, things are a little better than jvm: I mmap the IL program into 
memory, which gives me certain performance improvements when accessing method 
bytecode.  This isn't as easy with jvm due to the structure of the .class file.
<carnow> hey
<t3rmin4t0r> the static methods in glibc need to be loaded only once
<absurd> does the jvm share static methods?
<S11001001> carnow: what's hacking on pnetlib?
<S11001001> absurd: no
<rhysw> I'm confused - glibc gets most of its sharing from ELF/PIC - the code 
segment is read-only and doesn't need to be modified to load it.  This allows 
the kernel to optimise multiple processes that use the same .so.  But data 
sections must be per-process.  The more data, the bigger the load on the 
system.  JVM and IL both have the problem that most of what they need at 
runtime is data, not read-only code.
<carnow> haven't been doing much on pnetlib
<carnow> I was working on Socket, when someone else said they were working on 
oit
<S11001001> rhysw: I am not talking about the state. the program code is 
independent of the state, the state is on the stack
<rhysw> I can optimise the loader a bit, but intrinsic design weaknesses in IL 
and JVM limit what I can do.
<t3rmin4t0r> carnow: mix your codes ;)
<rhysw> The metadata contains a flattened version of the class definitions.  
This must be converted into a different data structure to allow the engine to 
process it.  This takes up heap space.  The more space, the slower loading 
becomes.
<roybatty> i'm getting some errors in cscc/common about some header files not 
being found
<roybatty> just co'd from cvs
<rhysw> what are their names?
<S11001001> "JVM and IL both have the problem that most of what they need at 
runtime is data, not read-only code." exactly...but the program code is still 
separate from the data it manipulates, and you are just regenerating the same 
thing every invocation
<roybatty> cg_nodes.tc:28: cg_defs.h: No such file or directory
<roybatty> cg_nodes.tc:35: cg_gen.h: No such file or directory
<roybatty> cg_nodes.tc:36: cg_output.h: No such file or directory
<roybatty> cg_nodes.tc:37: cg_utils.h: No such file or directory
<roybatty> and jv_nodes.tc:29: jv_output.h: No such file or directory
<rhysw> wtf?  Those files are in "codegen", where they've always been.
<S11001001> this is all really theoretical, and I have no illusions about it 
actually happening...but I just think it would be nice of there could be shared 
codespace
<S11001001> s/of/if/
<roybatty> hmm
<roybatty> yep, their there
<rhysw> S11001001: it would be nice, but the design of IL/JVM precludes it.
<S11001001> aiieee
<S11001001> well, I'll just quit now then. My participation in DotGNU rested on 
this becoming a reality ;)
<absurd> hmmz. i wish i could understand half of the above 20 lines
<S11001001> has anyone tracked down ajmitch?
<31mù> SachaS [ddress@hidden;30m] has joined #dotgnu
<SachaS> +
<fitzix> hmm... that was wierd... pnet configure script couldn't auto determine 
my host type
<rhysw> roybatty: any joy yet?  Perhaps a "make distclean" followed by a clean 
configure and re-make may help?
<fitzix> I fed it to it manually and that worked...
<roybatty> rhysw: I'll try it
<rhysw> fitzix: what OS are you running?  My "config.guess" could be out of 
date.
<fitzix> Red Hat 7.2
<S11001001> --host=i386-unknown-linux-gnu?
<fitzix> that works, as does i686-linux-gnu
<31mù> SachaS ddress@hidden;30m] has left #dotgnu 
["Client Exiting"]
<rhysw> that's weird - I've got either 7.1 or 7.2 on my primary machine (I 
forget which)
<31mù> SignOff carnow: #dotgnu ("BitchX: the 
quilted quicker picker upper")
;33mS11001001/#dotgnu runs slackware 7.1 with a bunch of 8.0 and 
slackware-current packages
<absurd> i run slackware 8
<fitzix> extra info:
<absurd> does slackware support rpm? i forgot
<fitzix> I have gcc-2.95.3 located at /opt/gcc
<fitzix> the host target there is i686-pc-linux-gnu
<S11001001> I didn't know a compiler was optional on a gnu system :)
<fitzix> everything's optional except for a kernel, various system utils, and 
bash :)
<fitzix> brb -- and compiling
<absurd> why, even osX comes with gcc
<31mù> gopz [ddress@hidden;30m] has joined #dotgnu
<gopz> again, my power on the blink 
<S11001001> fitzix: forget the system utils...use init and bash :)
<fitzix> S11001001: Actually, that really is pretty much all you need... I can 
actually write ls as a bash shell script
<fitzix> some of the others would be a bit difficult/impossible though
<S11001001> but seriously, I advise everyone who asks about 'linux' to make 
sure to install the development environment, regardless of programming skill
<rhysw> makes it easier to fix problems for them when they ring up screaming in 
the middle of the night :)
<fitzix> S11001001: Same here -- I'm an exception though :) -- and, I installed 
it in /opt so I wouldn't overwrite the standard distro compiler and gclibs... :)
<fitzix> rhysw: also, treecc didn't complain about my target host... I don't 
know if that's at all enlightening
<gopz> I think the attitude of DotGNU about bugs should be -- " If you run into 
any problems -- we'll fix it "
<S11001001> well, I am having that problem too, because I got the slackware 
2.95.3 pkg, and I don't want to overwrite 2.91.66 because of Linux
<fitzix> Well, that's generally my attitude and I think that that works for 
everyone else here as well... but they'd have to be actual bugs in the code, of 
course
;33mrhysw/#dotgnu thinks that people have config.guess problems when they 
can't figure out what compiler they want anyway probably shouldn't be 
complaining :)
<rhysw> s/people have/people that have/
<fitzix> Heh - I know what compiler I want :)
<S11001001> I want the standard compiler for HPUX
<fitzix> I just also know that Red Hat acts funky if you replace the original 
gcc libs :)
<fitzix> and I like to be able to delete things on the spot.. I love using /opt
<S11001001> fitzix: try --with-cc=...there's some configure option
<S11001001> fitzix: I don't think it's with-cc tho
<rhysw> CC=blahcc ./configure
<fitzix> no need - I have $CC set
<gopz> rhysw: I've got the pnetcgi demo up on my box ;)
<fitzix> hmm.. I'm missing those files (cg headers) as well
<fitzix> gcc -DHAVE_CONFIG_H -I. -I. -I../../include -Wall    -g -O2 -c 
cc_errors.c
<fitzix> In file included from cc_errors.h:24,
<fitzix>                  from cc_errors.c:21:
<fitzix> cg_nodes.tc:28: cg_defs.h: No such file or directory
<fitzix> cg_nodes.tc:35: cg_gen.h: No such file or directory
<fitzix> cg_nodes.tc:36: cg_output.h: No such file or directory
<S11001001> yeah, I just remembered that
<fitzix> cg_nodes.tc:37: cg_utils.h: No such file or directory
<rhysw> gopz: woohoo!
<fitzix> In file included from cc_errors.h:24,
<fitzix>                  from cc_errors.c:21:
<gopz> http://203.200.148.247/cgi-bin/pnetcgi
<gopz> for now (until my next power failure)
<S11001001> I did cd /usr/local/src/Python-2.1.1; ./configure --help | less, 
then realised that env | grep CC would have worked better :)
<rhysw> I'm investigating ...
<roybatty> fitzix: yep. that's what i got
<31mù> SignOff nymia: #dotgnu (Read error: 110 
(Connection timed out))
;33mS11001001/#dotgnu floats away for a while to try and write a summary of 
session 4
<gopz> S11001001: where are the logs ? -- can you put them on DotGNU.org ?
<S11001001> the logs I have have lots of bitchx crap in them...I'll upload them 
to geocities and offer a link, but I don't think posting them is a good idea
<31mù> SignOff t3rmin4t0r: #dotgnu (Read error: 
110 (Connection timed out))
<absurd> wow. i didn't know that richard stallman was so hairy!
<fitzix> is the cg_nodes stuff new?
<absurd> christ. he looks more like a GNU then a human!
<31mù> gopz is now known as t3rmin4t0r
<absurd> S11001001, want me to help you get rid of the BX stuff?
<fitzix> hmm.. it trips on all of the includes in cg_nodes.tc except for 
cg_nodes.h
<fitzix> ooh
<fitzix> I found the problem
<fitzix> rhysw: The directory... it's in <src>/cscc/common when it tries to 
reference cg_nodes.tc
<rhysw> I hope you have, because I just got a clean build from a fresh cvs 
checkout
<fitzix> all of the header references in cg_nodes.tc are to current directories
<fitzix> if you copy - say cg_defs.h into <src>/cscc/common it doesn't complain
<rhysw> then there is something seriously 100% broken about your compiler - 
because it shouldn't be doing that.
<roybatty> rhysw: I have the same problem
<roybatty> rhysw: exact same
<fitzix> hmm...
<absurd> S11001001, I managed to cut out the jargon!!!
<fitzix> roybatty: Did you update your gcc to gcc-2.95.3?
<roybatty> doing another build now
<roybatty> after a make distclean
<roybatty> i'm 2.95.4
<t3rmin4t0r> ahh !, I got the same error . Seems like you put an extra ../ on 
the include ?
<fitzix> rhysw: which gcc do you use?
<fitzix> hmm...
<t3rmin4t0r> sorry, the problems seems to be the build order of the files
<rhysw> gcc version 2.96 20000731 (Red Hat Linux 7.1 2.96-81)
<fitzix> hmm.. which is actually pre-2.95.3
;33mS11001001/#dotgnu is going to uninstall 2.91.66 and just install 
2.95.3...screw linux
<fitzix> err.. pre as in before
<31mù> SignOff kn4l: #dotgnu ()
<t3rmin4t0r> make in the codegen directory and re run make in compiler ?
<t3rmin4t0r> rhysw: is the cg_nodes.h generated by treecc ?
<rhysw> I still don't see what the problem is: #include "xxx.h" should look in 
the same directory as the file it is included from prior to looking in the 
current directory.
<rhysw> t3rmin4t0r: yes
<t3rmin4t0r> rhysw: the cscc is built before codegen ???? 
<fitzix> rhysw: I'm taking out my egcs RPMs and seeing if that changes anything
<rhysw> nope - codegen is first in the global make
<t3rmin4t0r> codegen is not the first in my make ????
<31mù> SignOff S11001001: #dotgnu ("codegen is 
not the first in my make ????")
<t3rmin4t0r> it compiles lots of other stuff (ildiff/image) etc before codegegen
<rhysw> SUBDIRS = libffi libgc include support image dumpasm engine ilasm 
ildasm \
<rhysw>           ilalink ilsize ilnative ilfind ildiff codegen cscc resgen 
ildb \
<rhysw>           csdoc csant samples doc tests
<fitzix> mine compiles lots of other stuff first as well
<31mù> nymia [ddress@hidden;30m] has joined #dotgnu
<roybatty> same here
<rhysw> there does seem to be some kind of treecc thing going on, but I still 
can't see how it accounts for the mis-build.
<t3rmin4t0r> rhysw: do these people forget the compile treecc first ?
<absurd> S11001001, you around?
<t3rmin4t0r> absurd: no , he left to write the summary
<rhysw> t3rmin4t0r: I don't think that is it
<31mù> S11001001 [ddress@hidden;30m] has joined 
#dotgnu
<t3rmin4t0r> rhysw: I mean export TREECC=../treecc stuff ?
<S11001001> geez...I uninstalled egcs pkg, only to find it had, for some 
idiotic reason, libstdc++ in it, and KDE crashed
<absurd> S11001001
<31mù> belrick [ddress@hidden;30m] has joined #dotgnu
<S11001001> only Mozilla kept working ;)
<absurd> I am able to fix up the logs so they look right
<S11001001> absurd: if you can do that, please send
<t3rmin4t0r> rhysw: I got a good compile !
<rhysw> t3rmin4t0r: what did you do?
<S11001001> BTW, session 5 is now over!
Log Ended at Sun Mar 17 0610GMT