[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [xougen] Some thoughts about X....

From: Per Cederberg
Subject: Re: [xougen] Some thoughts about X....
Date: Mon, 13 Oct 2003 00:03:29 +0200

NOTE: This is a long boring post. Now, I've warned you.

On Sun, 2003-10-12 at 22:03, Stefan Klinger wrote:
> Well i finally done some thinking about X(ouvert)....
> Nearly everybody seems to like the idea of having X modularized, why 
> then not stay compatible with the X-Protocol and other basic X 
> interfaces, but write the actual implementation in an OOP Language like 
> C++ ?

Please... What problem are you trying to resolve here?
Wasting time on a rewrite of such a magnitude is just 
not sensible, no matter how much one likes whatever 
language or paradigm of the day.

> Also making the development of drivers and such things easier would 
> maybe help, that companies which don't want to share/open there hardware 
> documentations will write there own driver (not ideal to have an binary 
> driver but better than having no driver at all).

The driver model (and possibly documentation) in modern 
XFree86 is pretty good, at least judging from comments 
I've seen elsewhere. Please be specific if you'd like
to enlighten us on any particular problem.

XFree86 is quite monolithic, though, causing driver 
releases to be infrequent and bundled with the complete
X server. Fixing this is one of the main drivers behind
Xouvert (as I understand things).

> About the server side widgets - i think predefined widget sets will 
> never survive, but maybe it would be possible to "upload" a set of 
> predefined "drawing operations" and "pixmaps" or something like that. So 
> a toolkit can be written that it can upload more complicated widgets to 
> the server.

Uploading "drawing operations" sounds like a pretty 
complex architecture to design. Projects such as Y that 
are using server-side widgets all allow direct client 
drawing instead (just as X do). That is simple, fairly 
efficient and cause people to think again before 
creating their own fancy widgets.

> An other really actual problem seems the clipboard, there should be an 
> well know extensions that the clipboard will support mime content or any 
> other way that an applications knows what type of data it contains. Then 
> it would also be possible, that an application tells which data types it 
> supports and let automatic conversion happen.

Seems like this already exist. See
http://www.jwz.org/doc/x-cut-and-paste.html and the 
section "Extra Credit: Content Negotiation" for example.

> And an other important thing would be to have a better documentation 
> than X - maybe an seperate documentation project should be created in 
> paralell to Xouvert.

Better documentation is always popular to ask for. In 
practice, though, if you don't specify exactly what 
areas are unclear or unspecified you will not end up 
with anything useful, just more words on paper. 

If you do have an issue with some part of the X 
documentation, I suggest you mention it here and maybe 
people will get interested in helping you to improve 
that. But please research any complaint (i.e. check man 
pages and existing docs properly) before writing to the
list about it.

> Something I am not sure about are printers, aren't they very much like 
> screens. Wouldn't it be possible to have a printer server work a bit 
> like an X-Server, or just let it be a special kind of X-Server?

Some people thought they were, and designed Display
PostScript from the famous page description language for 
printers. That has since evolved into PDF (in a sense) 
and is now being used by Apple as part of their Quartz 
display system.

Now, they went from printers to screens as PostScript in 
general is vectorized, which is needed as printers have 
very different resolutions. X of today is still pretty 
much bitmap-based (in practical usage), although support 
for vector graphics is available in the Cairo extension 
(http://cairographics.org/). Thus, making printer servers 
more like X servers is not a very good idea. (Not to 
mention that printers generally are not very interactive, 
or that good printing solutions already exist.)

> If I just reviewed allready existing things or if sombody thinks my 
> ideas are total nonsens - please give your opinion.
> Stefan Klinger

Well, no offense, but the ideas in this mail were all
pretty worn-out. Now, I've made my own share of "stupid" 
questions, so don't be intimidated or discouraged by any 
of this. You might want to Google a bit or check the 
Xwin.org site before making further suggestions here, 



Per Cederberg, Software Consultant

reply via email to

[Prev in Thread] Current Thread [Next in Thread]