Re: Hal version for GNOME 2.17

[Richard Hughes <hughsient gmail com>]

On Tue, 2006-09-19 at 18:50 +0200, Tomasz Torcz wrote:
>   Please, don't inflate GNOME dependences on non-technical merit. If
> some program can be compiled with library version X, it's NOT OK to
> depend on library version X+1 or X+2.
>   G-P-M is good example: its real dependency is HAL 0.5.6. Requiring
> higher version is unfair to users.

Umm no. I get frequent bug reports for gnome-power-manager that are
really bugs in HAL that were fixed *months* ago. I would go as far to
say that my bugs/week value would halve if there was a hard dep on
0.5.8. You really want to see the hal ChangeLog and look for "power" and
"battery" - there are lots of changes in the last year or so.

If you want gnome-power-manager to use CPU scaling support, with all the
enhancements and bugfixes, then the "real dependency" is 0.5.8.

For HEAD, I've just upped the gtk+ requirement to 2.10 and dropped
eggtrayicon support. Why? Because of the bugs in eggtrayicon and the
better solution of GtkStatusIcon. That's pretty aggressive (you need a
development distro, jhbuild or garnome to build the HEAD at the moment)
but relying on multiple layers of #ifdefs is really bad for testing.

On that matter, there's about 50 lines of obsolete code in various
classes of g-p-m that can be removed if the HAL dep is increased.

My opinion is to aim to use 0.5.8.[x] for 2-17 as by the release date of
GNOME 2.18, most of the distros will be releasing > 0.5.6 version of
HAL.

It's not fair to ask of maintainers "add GtkStatusIcon enhancements but
also keep the ~15 #ifdefs around for users of eggtrayicon" even though
stable distros are not going to release gtk+ 2.10 package updates.

Also, Elijah mentioned about the build issues... Anyone got any links to
specific problems?. Thanks.

Richard.

p.s. Sorry if that read like a rant, it has been a _long_ day! :-)