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

Re: Xdg-list digest, Vol 1 #432 - 11 msgs



On Wed, Jul 23, 2003 at 01:19:36AM -0500, Linas Vepstas wrote:
> On Wed, Jul 23, 2003 at 07:50:10AM +0200, Martin Konold was heard to remark:
> > 
> > Basically xdg is more about common standards for interoperability and making 
> > it easier for vendors like RH to "unify" the desktop offerings than about 
> > sharing code.
> 
> yeah, I'm sorry I wrote that last note.  Shouldn't have.
> 
> > Trying to stuff code down the throats of people not interested is 
> > contra-productive and does harm to the whole idea of freedesktop.org.
> 
> Yes. So, why exactly, were we talking about glib anyway? 

I think people may be taking things in a too stringent manner here.

XDG should produce specs, then perhaps it can also distribute a reference
'platform' set of libraries and code.  That doesn't really mean that
everybody has to use that code.  But those that want, can.  I don't see any
problem in having say two implementations part of the same 'platform' either.
Say one C++ and one C impl, or whatnot.

I think what havoc was trying to propose is that there be this set of code
that implements the various standards, and if you wish to follow the spec
while not writing a lot of code, you can just depend on this set of
libraries.

Just like no one forces anyone to use libpng, or openssl, but most people
tend to find it easier to just use those libs rather then redo the work
to implement those specs.

Some things, such as the libs jody mentioned for document importers may have
specs that are WAY too complex to implement over again.  And if the desktops
want to realy be interoperable, they should share that lib.  If the spec is
very complex, or not well defined (such as ms office formats), two
implementations are likely to have some different behaviour, that may make
interoperability much worse.  I mean if MS used the same libs for document
reading that we do, then we'd be 100% interoperable.

Personally I'm very pragmatic, and if a lib uses STL and I want to use it,
I'd go and write wrappers for C and wouldn't whine about it.  As long as
the lib is generally available for desktops and doesn't have random other
side effects when pulled in.

We're here to produce free software desktop.  Not everything will be as you
imagine it should be.  Most code is absolute crap.  GNOME, KDE, it's all
mostly crap.  Yet overall it generally works.  If we just wrote beautiful
code that is all theoretically correct and only uses nice well designed APIs,
we wouldn't have shit.  There are projects that are trying to do that,
usually they start redesigning the kernel first, and after about 10 years
they're still nowhere (many such projects come to mind).  Unix APIs are
utter shit.  Yet we semi hide them behind slightly better looking APIs and
the Unix semantics tend to bite us less often that way.  Yet somehow we
go on without rewriting the kernel.  Same with X11.  We bite the bullet and
use it ...  Let's produce free software desktops that work, rather then
lots of beautiful code that will never be finished.

George (The other George)

-- 
George <jirka 5z com>
   How shall we fuck off, o Lord?
                       -- Monty Python




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