* Micah Dowty ([EMAIL PROTECTED]) wrote:
> That's a good point. Another important distinction is that many of the
> apps in there are either broken, or just not in general useful to
> non-developers.
> 
> I propose that the more important apps, like pgl, battleship, blackout,
> pgboard, atomicnav, gridgame, etc. be included in their own packages,
> and that the apps that don't really do anything much like "demo" and
> "canvastst" are included in a picogui-demos package or something.

Hmm.  On one hand, I can see how eventually when picogui is a
production-level system, people may want certain apps installed and not
others (especially if disk space is limited).

On the other hand, the entire picogui-apps-cvs .deb is 119k at the
moment.  I can hardly justify breaking a 119k package into more than 5
smaller packages. The benefit is very small, at most 100k of bandwidth
and disk space per pgui user, and only on the offchance that someone
knows they want atomicnav but not gridgame. The cost is large, every
extra package increases the size of the Packages.gz file that *every*
Debian user downloads on *every* update, in addition to extra processing
time and disk/memory use for the clients.  The number of packages in
Debian grows very fast already, and I don't want to gratuitously
accelerate that rate.

For precedent, look at the xbase-clients package.  It contains all the
miscellaneous clients distributed with xfree86 -- 120 of them, for a total
of 1.5M!

On Thu, Feb 21, 2002 at 05:32:38PM -0300, Lalo Martins wrote:
> In the end, you could also have a "task-picogui" metapackage
> which installs the whole paella.

The task- prefix is no longer in vogue.  The metapackage would probably
be called picogui or picogui-system.

-- 
Joshua Haberman  <[EMAIL PROTECTED]>

_______________________________________________
Pgui-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/pgui-devel

Reply via email to