only question regarding point 2) I have is: do we really need PL/PMC to
be involved, Yes, definitely.
First, it makes absolutely no sense to use lowered rules than for RC2.
don't build anything from the sources"
is wrong. IBM does, and other people/companies might do as well.
If you get into
a blocker with JDT Core fixes, just ping me.
Thanks Dani, I want to clarify
the rules once again, may be I will put it somewhere in the wiki once we
agreed. So in order to push
commit to R4_XX_maintenance branch, one must follow RC2 criteria, which
are: "All fixes submitted
must have a project lead or PMC vote on the bug report, and the fix must
be reviewed by an additional committer (any committer other than the one
who made the fix). A positive review from a project lead or PMC member
means implicit approval and no vote is needed on the bug report." I read it this way: 1) On gerrit with
backported patch ask for the "+1" review from or from PL/PMC
or from a committer that was not the original patch owner. The committer who
backports the patch but is not the owner of the original patch, is also
allowed to vote with "+1". 2) If the "+1"
reviewer on gerrit is not PL/PMC, ask them in a comment on original bug
for an explicit "+1" for backporting. 3) If 1) and 2) are
successfully done, and tests are OK, gerrit can be merged. 4) API changes or
new features backported require additionally explicit PMC approval. Is this above right
so far? The only question
regarding point 2) I have is: do we really need PL/PMC to be involved,
even if we don't build anything from the sources and in theory nobody is
at risk? The PL/PMC vote is
worrying me, looking on the number of JDT core fixes I've backported so
far on 4.12 branch and the known lack of resources in JDT team. Kind regards, Andrey Loskutov
Спасение утопающих - дело рук самих утопающих
https://www.eclipse.org/user/aloskutov Gesendet: Freitag,
28. Februar 2020 um 14:54 Uhr Von: "Daniel Megert" <daniel_megert@xxxxxxxxxx> An: "Eclipse platform release engineering list." <platform-releng-dev@xxxxxxxxxxx>,
"Eclipse platform general developers list." <platform-dev@xxxxxxxxxxx> Betreff: Re: [platform-releng-dev] 4.15 maintenance branch Hi Andrey Thanks for bringing this up. Since the release train moved to a quarterly release cycle there was zero
interest to contribute to the maintenance branches that get created when
moving to the next release. It was only used by IBM and fed with backports
by IBMers (maybe with some exceptions that I can't remember right now).
Right from the beginning IBM wanted to make those fixes available for everyone
and not keep them in a private branch. It would be great if other people
use the maintenance branches (also old ones) to backport critical fixes,
or fixes that are important to their work/product/company. Given that the code is merged after RC2, the same rules must be followed
as for RC2 in https://www.eclipse.org/eclipse/development/plans/freeze_plan_4_15.php.
This also means: no API changes or feature work without PMC approval. The Platform Releng team will not spend resources to do maintenance builds.
However, anyone can use the source and build it but must not expect the
Platform Releng team to hold hands. So, welcome Andrey! Dani
Loskutov <loskutov@xxxxxx> To: "Eclipse
platform release engineering list." <platform-releng-dev@xxxxxxxxxxx>,
"Eclipse platform general developers list." <platform-dev@xxxxxxxxxxx> Date: 27.02.2020
21:03 Subject: [EXTERNAL]
[platform-releng-dev] 4.15 maintenance branch Sent by: platform-releng-dev-bounces@xxxxxxxxxxx
since we've started with Eclipse quarterly release cadence we've basically stopped development on maintenance branch. I saw that few commits were done here and there, but I didn't found any strategy / guidance.
In short, I maintain private 4.12 SDK branch with lot of backported patches from master. I do this for our product, and I plan to start another maintenance branch for 4.15 soon.
This time I believe it would make sense to not maintain my own private branch, but just push backported patches we need to the "official" R4_15_maintenance branch.
What would be the "right" way to do so? Do I need any approval
from component leads, do I need extra bugs opened to backport regression fixes, any "+1" from other developers? Or can I just push, run
gerrit build, and if it is green, merge? Since no one seem to work on maintenance branches anyway, this should be not a problem for anyone?
Another question would be if we could produce some regular SDK snapshot builds from that branch, and if there are resources & interest for
this? I personally don't need that build, we build SDK locally, but it could be something people are searching for.
I'm asking here because I believe it makes sense to give results of my (paid) maintenance work back to community, and it is really sad to see people searching for some fixes on a "known good" SDK version
but seeing only the way to get a new major SDK version with new unknown regressions instead.
Please note, I do not plan to maintain backports branch for every major SDK release, I have only resources (and plan) to do this for *some* releases.
So there are basically two questions:
1) What are the rules for backporting patches to maintenance branch? 2) Any interest in producing SDK builds from this branch?