Proposed: libpcre

[Havoc Pennington]

On Sun, 2004-08-01 at 02:46, Brad Hards wrote:
> On Sun, 1 Aug 2004 15:18 pm, Havoc Pennington wrote:
> > Maybe this points out that the freedesktop.org platform should be
> > limited to things that need to be shared for interoperability and/or
> > consistent user experience across apps.
> Serious question: what does "consistent user experience mean"?

The various UI guidelines out there are full of examples:
http://developer.gnome.org/projects/gup/hig/

> > If app A uses one regexp library and app B uses another, that really
> > doesn't impact either interoperability or user experience, other than
> > the fact that one regexp lib may be higher or lower quality.
> Does slightly different regexp syntax meet the needs of a consistent user 
> experience?

Not really, since no reasonable desktop UI has regexps in it ;-)
(Yes I realize some things the desktops ship have them...)

> > I don't think freedesktop.org should be in the business of blessing one
> > implementation detail vs. another. The point here is interoperability
> > and avoiding the negative user impact you get from e.g. multiple MIME
> > systems.
> I think that this is a dangerous line of reasoning. In the extreme case, why 
> do you want any X libraries? 

I'm not really interested in reductio ad absurdum, clearly all the lines
are blurry. I just think it's a mistake to be pulling in things that are
only very loosely desktop-related. Also, the first release of
freedesktop.org should be very conservative about how much to include.

> > But there's another line that may be relevant: fd.org platform should
> > potentially only include packages that are actively part of the fd.org
> > community. So e.g. this would exclude libpng.
> Who is the target audience for the fd.o platform? Who is going to install it? 
> Who is going to use it as a prerequisite? If you think it is for developers, 
> we probably need to make sure it meets developers needs.

My definition is that it's a "substrate for platforms" or "backend for
toolkits" - in other words it should basically contain the *shared*
dependencies of Qt, GTK+, XUL, WINE and other common desktop toolkits.
Then we should add the window manager stuff since the WM spec is
effectively part of providing toolkit functionality.

In other words, I'm often not expecting anyone to code to
freedesktop.org libraries directly, or exclusively; the fact that we
won't include a toolkit is the most obvious example.

But I also don't think we should include things like an ODBC-style
database layer, a scripting language, an IDE, or whatever. Let's stick
to our core competency.

http://ometer.com/DesktopCon2004.sxi has some notes I presented at
DesktopCon

Havoc