There are a large number of project that have abandoned the '-devel/-stable'
process in favor of ongoing introduction of new features into the mainline (the
kernel, firefox, and chrome are all examples of this that work well). So I
think it's a very reasonable thing to do. If you have a major change (like the
changes that global variables triggered between v7 and v8), then creating a
temporary -devel tree for people to test that then gets abandoned when the
feature gets merged into the mainline is a reasonable thing to do.
With the change from the develop-stabilize-release approach, I would also
suggest thinking hard about doing a time-based release approach. Every 6 months
or so (planned to fit reasonably with distro release schedules) release a new
8.x with all changes in the meantime going into 8.x.y.
thoughts?
On Wed, 29 Oct 2014, Rainer Gerhards wrote:
Hi all,
it may sound strange, but I strongly think about dropping -devel versions
and instead moving new features directly into the -stable branch.
The reason is that almost nobody nowadays tries out the -devel versions.
The past two years, I've always seen the same pattern: when I started a new
-stable branch, a lot of bug reports immediately appeared - bugs that
obviously were not detected because nobody used -devel. The really bad
thing about this is that usually the feature causing the bug was
implemented some month ago, so I do not have a clear memory what may be the
root cause. Also, in a new stable branch there are many changes intermixed,
which makes troubleshooting even harder.
As such, I consider a policy change where we will support the current and
previous stable release (right now that would be 8.4.2 and 8.4.1) and
enhancements going directly into the -stable release. Actually, we would
drop the -stable, -devel qualifiers, it would just be "the rsyslog v8
release".
Let's consider the next version: changes would go into 8.4.3, but we would
still support 8.4.2 in regard to questions. So if someone hits a regression
with 8.4.3, he would need to go back to 8.4.2 until 8.4.4 is released.
On the plus side, that would also mean new features would be more readily
available, in contrast to the 3 to 8 month wait period we currently have
for those that insist on stable versions.
I am not sure, however, if we should release new versions more rapidly than
we did with -stable versions. Technically, it makes sense, but many users
don't like that (I know from past conversations).
Comments appreciated.
Rainer
_______________________________________________
rsyslog mailing list
http://lists.adiscon.net/mailman/listinfo/rsyslog
http://www.rsyslog.com/professional-services/
What's up with rsyslog? Follow https://twitter.com/rgerhards
NOTE WELL: This is a PUBLIC mailing list, posts are ARCHIVED by a myriad of
sites beyond our control. PLEASE UNSUBSCRIBE and DO NOT POST if you DON'T LIKE
THAT.
_______________________________________________
rsyslog mailing list
http://lists.adiscon.net/mailman/listinfo/rsyslog
http://www.rsyslog.com/professional-services/
What's up with rsyslog? Follow https://twitter.com/rgerhards
NOTE WELL: This is a PUBLIC mailing list, posts are ARCHIVED by a myriad of
sites beyond our control. PLEASE UNSUBSCRIBE and DO NOT POST if you DON'T LIKE
THAT.
