25 Feb
2012
25 Feb
'12
12:12 p.m.
The only time I've ever seen this work is with OpenBSD - but that's
because they do a strict six-monthly release schedule, with set periods
of time hacking away, set periods of time testing things, and set
periods of time finalizing the release. Any new features that doesn't
make it into the (I think it's 4 months?) hacking period just doesn't
make it. In other words, auditing is very strict there; that's hardly
what I could say for other projects. And fair enough; it's a huge job!
And some of the big rewrites aren't even enabled/in the main tree until
years later!
Quick releases work the way you are talking about *in theory* - while in
practice (at least, here in ReactOS that is) I have noticed a lot of "Oh
we're waaaay past the deadline dude. Let's like, you know, leave this
big regression till the next release, man." - in other words, great in
theory, but rarely practical and benefits are either negative, or
negligible.
The MM rewrite was an example. Now that's been pushed to the next
release. (correct me if I am wrong) - it was going to be released with
0.PI before.
On Sat, 25 Feb 2012 11:57:57 +0000
Andrew Faulds <ajfweb(a)googlemail.com> wrote:
Quicker releases have the benefit that it forces us to
fix lingering
regressions and release. Otherwise regressions and significant bugs
pile up until release and we have to spend months fixing them.
On 25 February 2012 04:00, Adam <geekdundee(a)gmail.com> wrote:
"Faster release cycle is better" is
what a bunch of idiots at
Mozilla Corporation said.
And now their browser has begun the process of becoming more
bloated, much slower, and contains more and more useless rubbish and
"features" with every release, while the competition (including
Internet Explorer) is excelling on simple and useful things, such
as a UI that doesn't randomly freeze solid every five minutes.
I strongly oppose this "faster release cycle" unless it means simply
tagging a TRUNK build that "just works" as a release and releasing
without doing anything further, apart from maybe the occasional
bootscreen/login screen change to reflect the version number.
If you want to try new features, use a trunk build.
Quick release may actually *reduce* the interest in ReactOS when
people try the "stable" builds only to find that it is a lot
buggier because of the absurdly limiting deadlines, and so few
final fixes could be made.
On Fri, 24 Feb 2012 16:31:52 +0000
Andrew Faulds <ajfweb(a)googlemail.com> wrote:
Good! Let's proceed with 0.3.15... a faster
release cycle is
better, and will help interest in ROS as people will be able to
see clearly that progress is being made (as opposed to appearing
to be stagnant to those who do not look at trunk).
Plus, I want a "stable" release to try some new features ;P
On 24 February 2012 11:49, Amine Khaldi <amine.khaldi(a)reactos.org>
wrote:
_______________________________________________
Ros-dev mailing list
Ros-dev(a)reactos.org
http://www.reactos.org/mailman/listinfo/ros-dev February 2012 Meeting Minutes
2012-02-23
19:45 UTC
Fezile, #meeting
Proceedings
===========
* Meeting started at 19:45 UTC by Aleksey Bragin.
* Point 1: Release status: 0.3.15
---------------------------------
* Aleksey Bragin proposed that we prepare trunk and release
0.3.15, as it is now, before anything major happens. Several
members then wondered about the status of USB and the "mshtml
bug". Jerome Gardou explained a bit about how the mshtml bug is
far from completely being fixed, because if you look at
testbot's results, there are still some problems with ASSERTs
hit and some bad pagefaults happening when paging out.
* A lengthy discussion occurred, about how ready trunk is,
whether it's in a better state than 0.3.14, the state of the
theme to be bundled with ros, the plan for CLT, whether we
should go 0.4<something> or 0.3.15... This was settled through a
voting: "Do you agree to release 0.3.15 with current trunk
features, before CLT?".
* The total number of votes was 21, with 11 votes as Yes, 5 as
No, and 5 abstentions. As a result, we will be having a 0.3.15
release with CLT being the deadline.
* Point 2: New website status and migration plans
-------------------------------------------------
* Amine Khaldi gave a quick summary on the state of the website
revamp:
- Quite some progress has been made in the theming department,
and it's visible from the playground. It's starting to look a lot
like the current one, which means it's almost done (only some
issues are left).
- Maciej Bialas has been investigating how to import users from
the RosCMS. We will most likely need to compromise: members will
have to reenter passwords when we migrate, but we'll see.
- Amine Khaldi then mentioned that we need help from web devs
familiar with Drupal. Alexander Rechitskiy mentioned that he has
a guy that wants to be part of the web team, and he passes his
info to Amine to establish the contact.
- He also mentioned that Maciej is very busy, so the progress
will slow down, and that the current short term plan is to
continue the theming work, provide a way to import users, import
the rest of data from wiki/bugzilla...etc and finally add the
phpbb bridge.
* He then summarized the summary by saying: We needs skillful
drupal guys, we're almost there, and that's it.
* Point 3: CMake migration and finally abandoning RBuild
--------------------------------------------------------
* Amine Khaldi suggested that it's time to ditch rbuild, as he
suspects most of our members have migrated already. Agreement on
that, from ours members, was unprecedented as *everyone* were on
favor of this :)
* We agreed on doing the necessary commits/infrastructure changes
ASAP, ie "tonight" (relative to the meeting of course), but that
didn't happen just yet so it's just a matter of time. Amine
Khaldi wants to do the honors (the commit that will remove
rbuild and its related files).
* Amine Khaldi also mentioned that he's preparing a nice little
surprise for major build performance boost, on many levels, so
stay tuned guys ;)
* Jerome Gardou explained that he'll be handling the PCH support
to get it to a much better state than it is right now.
* Point 4: Added per Amine Khaldi's request: GSoC preparations
--------------------------------------------------------------
* Amine Khaldi explained that he created a Google Doc and invited
the interested members. He asked the members to collaborate and
shape up an excellent ideas list, projects that fit in GSoC at
both the time and complexity levels.
* Colin suggested building up on the last year in terms of
informative content (wiki pages from last year), covering
questions about how we deal with students, our application
form... etc.
* Meeting closed at 21:45 UTC by Aleksey Bragin.
* Minutes written by Amine Khaldi.
_______________________________________________
Ros-dev mailing list
Ros-dev(a)reactos.org
http://www.reactos.org/mailman/listinfo/ros-dev