Re: [Help-nano] RE: compiling nano on HP-UX

[Chris Allegretta]

Hey Jay, did you get a chance to test out the 3rd build I posted below?  
Unfortunately I don't have access to an HPUX machine currently to try this out 
myself.  Thanks much.

Chris Allegretta <address@hidden> wrote:

>Okay, I was just testing for the variable too early on.  I fixed that
>as well, the new version is at
>http://www.nano-editor.org/dist/test/nano-2.1.9-svn-xopenchk3.tar.gz,
>or we may be able to get away with a replacement script (which would
>probably be worth finding out since I think we're going to be going
>through several iterations). It is attached.
>
>On Sat, Feb 21, 2009 at 10:02 AM, Chris Allegretta <address@hidden> wrote:
>> Ok, please be patient.   It's difficult to debug a system I dont have
>> actual access to :)
>> I will look back through the log you sent and see if the variable
>> $CURSES_LIB_NAME
>> is just not getting set for some reason.   Incidentally, this system
>> does have a /usr/include/curses.h, right?
>>
>> In the meantime, if you want to comment out the lines to force the new
>> check to work by putting a # at the beginning of lines 6716 and 6838,
>> it would be interesting to see what it finds.
>>
>> On Sat, Feb 21, 2009 at 9:43 AM, Jay <address@hidden> wrote:
>>> Chris, It is still skipping the new code.
>>> config.log:
>>>
>>>
>>> configure:6645: checking for GNU-style word boundary regex support
>>> configure:6682: gcc -o conftest -g -O2   conftest.c  >&5
>>> configure:6685: $? = 0
>>> configure:6691: ./conftest
>>> configure:6694: $? = 1
>>> configure: program exited with status 1
>>> configure: failed program was:
>>> ..
>>> .. no line numbers reported in here..
>>> ..
>>> configure:6709: result: no
>>> configure:6842: checking whether to enable UTF-8 support
>>> configure:6855: result: auto
>>> configure:6858: checking whether to use slang
>>> configure:7749: result: no
>>> configure:7843: checking for getdelim
>>> configure:7899: gcc -o conftest -g -O2   conftest.c  >&5
>>> ld: Unsatisfied symbol "getdelim" in file /var/tmp//ccocXNPa.o
>>>
>>>
>>> The XOPEN code is around like 6716.
>>>
>>> Maybe soon I'll have the right versions of perl/autoconf/etc. and can try
>>> myself.
>>>
>>>  - Jay
>>>
>>>> Date: Sat, 21 Feb 2009 01:53:32 -0500
>>>> Subject: Re: [Help-nano] RE: compiling nano on HP-UX
>>>> From: address@hidden
>>>> To: address@hidden
>>>> CC: address@hidden
>>>>
>>>> Are you saying it bails out, or the code I added just doesn't execute?
>>>> If it's the latter it's just due to me leaving out a $. I created a
>>>> new tarball at
>>>> http://www.nano-editor.org/dist/test/nano-2.1.9-svn-xopenchk2.tar.gz,
>>>> please give it a go at your leisure.
>>>>
>>>> On Fri, Feb 20, 2009 at 11:37 AM, Jay <address@hidden> wrote:
>>>> > Trying again, without the attachments.
>>>> >
>>>> > - Jay
>>>> >
>>>> >
>>>> >
>>>> >
>>>> >
>>>> >
>>>> > ________________________________
>>>> > From: address@hidden
>>>> > To: address@hidden
>>>> > CC: address@hidden
>>>> > Subject: RE: [Help-nano] RE: compiling nano on HP-UX
>>>> > Date: Fri, 20 Feb 2009 16:37:18 +0000
>>>> >
>>>> > Hi Chris, it doesn't work.
>>>> > The relevant code in configure doesn't run.
>>>> >
>>>> > config.log:
>>>> >
>>>> > |
>>>> > | if (regcomp(&r, "\\>", REG_EXTENDED|REG_NOSUB))
>>>> > | return 1;
>>>> > | if (regexec(&r, "word boundary", nmatch, &pmatch, 0))
>>>> > | return 1;
>>>> > | return 0;
>>>> > | }
>>>> > configure:6709: result: no
>>>> > configure:6842: checking whether to enable UTF-8 support
>>>> > configure:6855: result: auto
>>>> > configure:6858: checking whether to use slang
>>>> > configure:7749: result: no
>>>> > configure:7843: checking for getdelim
>>>> > configure:7899: gcc -o conftest -O2 conftest.c >&5
>>>> > ld: Unsatisfied symbol "getdelim" in file /var/tmp//ccofn2wV.o
>>>> > 1 errors.
>>>> >
>>>> >
>>>> > Your code is around line 6716.
>>>> >
>>>> >
>>>> > Attached config.log and config.h.
>>>> >
>>>> > - Jay
>>>> >
>>>> >
>>>> >> Date: Thu, 19 Feb 2009 19:56:42 -0500
>>>> >> Subject: Re: [Help-nano] RE: compiling nano on HP-UX
>>>> >> From: address@hidden
>>>> >> To: address@hidden
>>>> >> CC: address@hidden
>>>> >>
>>>> >> Hi Jay,
>>>> >>
>>>> >> I would rather just put feature-specific checks in configure.ac
>>>> >> whenever possible. I took a crack at this, can you try the tarball at
>>>> >> http://www.nano-editor.org/dist/test/nano-2.1.9-svn-xopenchk.tar.gz
>>>> >> and let me know whether it works for you? Thanks much.
>>>> >>
>>>> >> On Wed, Feb 18, 2009 at 8:22 PM, Jay <address@hidden> wrote:
>>>> >> > Chris, no problem, no rush. It is working for me.
>>>> >> > tarball or svn should be ok.
>>>> >> > (I've never used command line svn, but have used cvs, so... and I
>>>> >> > don't
>>>> >> > have
>>>> >> > svn on this machine yet, but can always go through another..)
>>>> >> >
>>>> >> >
>>>> >> > If you are worried about:
>>>> >> >
>>>> >> >
>>>> >> > fips_rand.c:59:1: warning: "_XOPEN_SOURCE_EXTENDED" redefined
>>>> >> >
>>>> >> > http://rt.openssl.org/Ticket/Display.html?id=1026&user=guest&pass=guest
>>>> >> >
>>>> >> >
>>>> >> > You can either do
>>>> >> > #undef _XOPEN_SOURCE_EXTENDED
>>>> >> > #define _XOPEN_SOURCE_EXTENDED
>>>> >> > Are there values other than 1?
>>>> >> > Like defining it to a date/version-thing?
>>>> >> >
>>>> >> >
>>>> >> > or #ifndef _XOPEN_SOURCE_EXTENDED
>>>> >> > #define _XOPEN_SOURCE_EXTENDED
>>>> >> > #endif
>>>> >> >
>>>> >> >
>>>> >> > but having configure only #define it if it "helps" is also
>>>> >> > reasonable.
>>>> >> > Maybe there are other problems.
>>>> >> >
>>>> >> >
>>>> >> > Another thing you can do, a little icky maybe, but maybe nicer than
>>>> >> > adding
>>>> >> > configurey :)
>>>> >> > is isolating the code that needs it so as to minimize the risk. You
>>>> >> > know,
>>>> >> > little wrapper functions and extern consts, with some alternate
>>>> >> > names,
>>>> >> > that
>>>> >> > just wrap the "real" stuff.
>>>> >> > Perhaps that's not the best source organization though.
>>>> >> >
>>>> >> >
>>>> >> > stdarg.h you can either put in unconditionally, should be safe.
>>>> >> > This system has a K&R compiler but I think still has stdarg.h.
>>>> >> > As well, it need only go in under the same situation -- tied to the
>>>> >> > use
>>>> >> > of
>>>> >> > #define _XOPEN_SOURCE_EXTENDED.
>>>> >> >
>>>> >> >
>>>> >> > SIMILARY, I had some source yesterday (not nano, not HP-UX) that
>>>> >> > required
>>>> >> > #define _GNU_SOURCE or such.
>>>> >> > g++ predefined it, gcc did not. So I had to do that #undef to fix
>>>> >> > that
>>>> >> > very
>>>> >> > same warning.
>>>> >> > It was C code but I like to check that it compiles as C++ too.
>>>> >> >
>>>> >> >
>>>> >> > - Jay
>>>> >> >
>>>> >> >
>>>> >> >> Date: Wed, 18 Feb 2009 19:44:07 -0500
>>>> >> >> Subject: Re: [Help-nano] RE: compiling nano on HP-UX
>>>> >> >> From: address@hidden
>>>> >> >> To: address@hidden
>>>> >> >> CC: address@hidden
>>>> >> >>
>>>> >> >> Hi Jay,
>>>> >> >>
>>>> >> >> Thanks for this report. Unfortunately, from what i read about
>>>> >> >> _XOPEN_SOURCE_EXTENDED it sounds like it may change other behaviors
>>>> >> >> as
>>>> >> >> well by defining it. I think I will have to add a check to the
>>>> >> >> configure script to test whether the colors are available without
>>>> >> >> defining _XOPEN_SOURCE_EXTENDED, and then try it with the flag and
>>>> >> >> set
>>>> >> >> it properly if needed, and then if needed define it. This will
>>>> >> >> probably take a day or two, but are you willing to test out a
>>>> >> >> tarball
>>>> >> >> if I create one for you, or do you have subversion access?
>>>> >> >>
>>>> >> >> The stdarg stuff should be easy to check for as well via configure.
>>>> >> >>
>>>> >> >> On Tue, Feb 17, 2009 at 4:50 PM, Jay <address@hidden> wrote:
>>>> >> >> >
>>>> >> >> > The include is of stdarg.h with angle brackets.
>>>> >> >> > My email is getting munged, sorry.
>>>> >> >> >
>>>> >> >> > - Jay
>>>> >> >> >
>>>> >> >> >
>>>> >> >> >
>>>> >> >> > ----------------------------------------
>>>> >> >> >> From: address@hidden
>>>> >> >> >> To: address@hidden
>>>> >> >> >> Subject: FW: compiling nano on HP-UX
>>>> >> >> >> Date: Tue, 17 Feb 2009 15:34:44 +0000
>>>> >> >> >>
>>>> >> >> >>
>>>> >> >> >> [bounced there]
>>>> >> >> >>
>>>> >> >> >>
>>>> >> >> >>
>>>> >> >> >> From: jay.krell@
>>>> >> >> >> To: nano-devel@
>>>> >> >> >> Subject: compiling nano on HP-UX
>>>> >> >> >> Date: Tue, 17 Feb 2009 15:20:58 +0000
>>>> >> >> >>
>>>> >> >> >>
>>>> >> >> >>
>>>> >> >> >> nano-2.1.9
>>>> >> >> >>
>>>> >> >> >>
>>>> >> >> >> -bash-3.2$ gcc -v
>>>> >> >> >> Reading specs from
>>>> >> >> >> /usr/local/lib/gcc-lib/hppa2.0w-hp-hpux11.11/3.3.6/specs
>>>> >> >> >> Configured with: /src/gcc-3.3.6/configure -verbose
>>>> >> >> >> -with-as=/usr/local/bin/gas -with-gnu-as
>>>> >> >> >> Thread model: single
>>>> >> >> >> gcc version 3.3.6
>>>> >> >> >>
>>>> >> >> >>
>>>> >> >> >> -bash-3.2$ uname -a
>>>> >> >> >> HP-UX jkhpux1 B.11.11 U 9000/800 1595760578 unlimited-user
>>>> >> >> >> license
>>>> >> >> >> This is I think "11v1".
>>>> >> >> >>
>>>> >> >> >>
>>>> >> >> >> color.c: In function `color_init':
>>>> >> >> >> color.c:86: error: `COLOR_WHITE' undeclared (first use in this
>>>> >> >> >> function)
>>>> >> >> >> color.c:86: error: (Each undeclared identifier is reported only
>>>> >> >> >> once
>>>> >> >> >> color.c:86: error: for each function it appears in.)
>>>> >> >> >> color.c:93: error: `COLOR_BLACK' undeclared (first use in this
>>>> >> >> >> function)
>>>> >> >> >> gmake: *** [color.o] Error 1
>>>> >> >> >> -bash-3.2$ pwd
>>>> >> >> >> /src/nano-2.1.9/src
>>>> >> >> >> -bash-3.2$ cp color.c color.c.orig
>>>> >> >> >>
>>>> >> >> >>
>>>> >> >> >>
>>>> >> >> >> put #define _XOPEN_SOURCE_EXTENDED at the top of color.c
>>>> >> >> >> =>
>>>> >> >> >>
>>>> >> >> >>
>>>> >> >> >> /usr/local/lib/gcc-lib/hppa2.0w-hp-hpux11.11/3.3.6/include/curses.h:755:
>>>> >> >> >> error: syntax error
>>>> >> >> >> before "va_list"
>>>> >> >> >>
>>>> >> >> >>
>>>> >> >> >> put #include at the top of color.c
>>>> >> >> >> => works
>>>> >> >> >>
>>>> >> >> >>
>>>> >> >> >> I don't have GNU diff yet:
>>>> >> >> >> -bash-3.2$ diff color.c.orig color.c
>>>> >> >> >> 23a24,25
>>>> >> >> >>> #define _XOPEN_SOURCE_EXTENDED
>>>> >> >> >>> #include
>>>> >> >> >>
>>>> >> >> >>
>>>> >> >> >> Either order compiles, but this seems better.
>>>> >> >> >>
>>>> >> >> >>
>>>> >> >> >> rcfile.c: In function `color_to_short':
>>>> >> >> >> rcfile.c:543: error: `COLOR_GREEN' undeclared (first use in this
>>>> >> >> >> function)
>>>> >> >> >> rcfile.c:543: error: (Each undeclared identifier is reported only
>>>> >> >> >> once
>>>> >> >> >> rcfile.c:543: error: for each function it appears in.)
>>>> >> >> >> rcfile.c:545: error: `COLOR_RED' undeclared (first use in this
>>>> >> >> >> function)
>>>> >> >> >>
>>>> >> >> >> same thing.
>>>> >> >> >>
>>>> >> >> >> -bash-3.2$ diff rcfile.c.orig rcfile.c
>>>> >> >> >> 23a24,25
>>>> >> >> >>> #define _XOPEN_SOURCE_EXTENDED
>>>> >> >> >>> #include
>>>> >> >> >>
>>>> >> >> >>
>>>> >> >> >> and then it builds and works.
>>>> >> >> >>
>>>> >> >> >>
>>>> >> >> >> Thanks,
>>>> >> >> >> - Jay
>>>> >> >> >
>>>> >> >> > _______________________________________________
>>>> >> >> > Help-nano mailing list
>>>> >> >> > address@hidden
>>>> >> >> > http://lists.gnu.org/mailman/listinfo/help-nano
>>>> >> >> >
>>>> >> >
>>>> >
>>>> > _______________________________________________
>>>> > Help-nano mailing list
>>>> > address@hidden
>>>> > http://lists.gnu.org/mailman/listinfo/help-nano
>>>> >
>>>> >
>>>
>>> _______________________________________________
>>> Help-nano mailing list
>>> address@hidden
>>> http://lists.gnu.org/mailman/listinfo/help-nano
>>>
>>>
>>