Future of Ac... or even of PLD
Marcin Król
hawk w limanowa.net
Czw, 22 Mar 2007, 14:33:21 CET
Hello.
Since there are quite big chances that this time Ac will really hit
"stable" state I was thinking about what to do with it in the future.
Since not many developers are interested in maintaining Ac I'm 99% sure
that we will have problem with updates. I've done quick stats from
cvs-commit. There were 1051 commits to AC-branch since 1st of January
while HEAD received 6613 commits.
Stats for AC-branch:
glen: 551 commits
hawk: 172 commits
qboosh: 77 commits
megabajt: 39 commits
undefine: 29 commits
shadzik: 23 commits
ankry: 19 commits
arekm: 16 commits
twittner: 15 commits
baggins: 13 commits
charles: 10 commits
blues, mguevara: 9 commits
adamg: 7 commits
rotom, zawadaa, zbyniu: 6 commits
dzeus, grzegorz, sparky: 5 commits
tommat: 4 commits
blekot, wolverine: 3 commits
aredridel, matkor, prism, sls: 2 commits
areq, arvenil, darekr, deejay1, jasw, kosmo, krystian, lisu, pascalek,
radek, spider: 1 commit
Lets say that person making over 20 commits in 3 months is counted as
"active Ac developer". We have 6 active Ac developers then. IMO its not
enough for keeping Ac up to date and fixing bugs that most likely will
appear. Therefore I think that Ac will die 2 or 3 months after sticking
"stable" on it because of security holes and old, buggy versions of
various software. Or maybe it already died?
Th probably will not get stable enough in next 2-3 months to replace
production systems, so here we are back with idea of "always in
developement" way for PLD. It was discussed few times so I'll post just
some basic information: two trees, one marked as stable, second as
devel. Things from devel that will be stable enough are moved to stable
tree. From time to time or after some big changes a new snapshot is
being released for all those .iso images fans. Third tree may be
introduced for completly unstable stuff like gcc snapshots etc.
I will at least try such way of developement for Ac. That means some big
changes for Ac 2.1. First of all, clean upgrade from 2.0 probably will
not be possible. By clean upgrade I mean 'poldek --upgrade-dist' going
w/o errors.
I would like to see following changes in Ac 2.1:
New software, so called "hot stuff", but stable versions only. Most
requested ones are:
- newer glibc (death sentence for i386 architecture)
- gcc 4.1.x
- modular xorg 7.2
- gnome 2.18
- put your favourite app(s) here
Architecture changes: deletion of i386 due to glibc change, possible
deletion of i586, alpha, sparc. Rename amd64 to x86_64.
Why to delete i586? Because a year from now probably no one will use old
Pentium 1 or AMD K6/K6-2 processors. If someone is using such old
hardware he may get base i686 system (400-500 MHz, 128 RAM, <=10 gig
HDD) for 40-50 PLN (about 10 euros). Some people are even giving such
configs for free as it is just junk laying around for them.
Why to delete alpha, sparc? Those two are most problematic ones and
barely few people are using them. There were many situation where some
updates were nailed by failing alpha/sparc build and there were no
volunteers to fix them. If someone wants alpha/sparc, he may unoficially
maintain those architectures, but keeping them out of official support
will ensure that "architecture blockers" will not cause us problems anymore.
I realize that whole idea may die due to lack of developers interested
in maintaining the Th-like Ac. However I want to hear your opinions,
comments, suggestions.
M.
P.S. If you want to help Ac live, please, declare a list of specs that
you will maintain after stable release.
Więcej informacji o liście dyskusyjnej pld-discuss