Configuration files (Re: [PATCH] strtol() fails on 32 bit...)

[Pekka Paalanen]

On Mon, 30 Jan 2012 09:35:03 -0500
Kristian Høgsberg <krh at bitplanet.net> wrote:

> 2012/1/30 Pekka Paalanen <ppaalanen at gmail.com>:
> >
> > I'm sceptical about the whole idea of expressing an ARGB tuplet as a
> > single integer because of possible endianness issues.
> 
> There's no endian issue here: we parse the integer and shift down the
> a, r, g and b components.  If we accessed by casting to unsigned char

Ooh, right.

> * and indexing we'd be in trouble.  Representing colors as 8 digit hex
> numbers is at least pretty common (thanks to the web and css), but
> perhaps not intuitive.
> 
> > I understand it is convenient for now, and the config parser is not
> > supposed to grow into a full-featured config system. Should we consider
> > starting to use a proper configuration library, whatever that might be?
> >
> > Maybe it could be a newbie task to find and integrate one?
> > To add to Vignatti's blog list of tasks. ;-)
> 
> I'm carefully trying to keep weston neutral and free of dependencies.
> Mainly "political" dependencies that would put weston in one of the
> "open source silos", but I also don't want esoteric/little know
> libraries on the dependency list.  And while I'm on my soap box, code
> reuse is overrated.  Unless a library is solving a big problem for is,
> in the right way, it's not worth the cost of taking the extra
> dependency and shoe-horning your problem into somebody else solution.
> 
> Recommended reading:
> http://altdevblogaday.com/2011/02/15/data-structures-one-size-does-not-fit-all/

I take that as a "no". Seems like I accidentally did the right thing
with the matrix inversion then.


Thanks,
pq