SeaMonkey's Release process depends on CVS for the patcher-configs update. As per bug 1057920, we need to move away from CVS. RelEng-wise, the patcher-configs has to be moved to Mercurial (if it hasn't been done so... ) and the buildbotcustom code needs to be modified to reflect this change. This needs to be done prior to CVS' shutdown in 2015 Q1. (which is soon.)

Status? End of March is approaching quickly, when CVS will shut down (see also bug 1057920 comment 7)

Hi hal, Thanks for poking me on this. I haven't done anything yet on this as I was working on the release and buildbot code bustage. Will get this done next.

this is the easy part.

As Laura commented in bug 1057920, CVS is scheduled to be decommissioned by today. We can hold on a little while longer but we need to close out the SeaMonkey bugs. Edmund/Callek - Can you please work to get this bug resolved.

(In reply to Lawrence Mandel [:lmandel] (use needinfo) from comment #4) > As Laura commented in bug 1057920, CVS is scheduled to be decommissioned by > today. We can hold on a little while longer but we need to close out the > SeaMonkey bugs. > > Edmund/Callek - Can you please work to get this bug resolved. Per talk with Lawrence, while this is a blocker for SeaMonkey to be off CVS, there is a blocker to us testing/validating whatever we do here. And that is something that is also blocking our releases at present. Current Plan: * If we have that other blocker fixed before CVS is down, we'll do a release using the CVS process that exists, and work diligently to move off CVS in that regard * We'll declare this bug (here) no longer a blocker to CVS decom as long as we have at least 1-2 business days notice, that allows us to archive the required data for the move somewhere. * We'll archive the data and use that when it does come time for us to migrate.

I'm wondering if we need the following cvs repos converted to hg as well? mozilla/tools/patcher, mozilla/tools/patcher-configs mozilla/tools/release/MozBuild mozilla/tools/release/Bootstrap/Util.pm reason for the patcher one is because of the use of UPDATES_RELEASE_R18 tag that we use. Right now, it's just a straight push-the-current-contents-of-patcher tools to Callek's tools repo, but that doesn't include the UPDATES_RELEASE_R18 tag. Or do we just use the default tag?

Including the UPDATE_PACKAGING_R18 tag.

Comment on attachment 8593181 [details] [diff] [review] Adding all related patcher files to one repo (v1) [option #1] Actually... this is the wrong approach.

I'm wondering if it's a better idea if we put the patcher stuff into the hg.mo/SeaMonkey/patcher repo. The reason why I suggest this is because if I were to put the patcher files into Callek's tools repo, and tag it with UPDATE_PACKAGING_R18, during the patcher steps, the UPDATE_PACKAGING_R18 tag will be before all those SEAMONKEY_2_??_RELEASE tags and I'm concerned it might affect the process. i.e. in hg.mo/users/Callek_gmail.com/tools, <cset-latest> tagged with SEAMONKEY_<ver>_RELEASE . . <cset ###> tagged with UPDATE_PACKAGING_R18 so anything between cset-latest and cset ### during the patcher step is ignored.. In any event, I'll have two patches made. 1) if it's under hg.mo/SeaMonkey/patcher 2) if it's under hg.mo/users/Callek_gmail.com/tools/release/updates/patcher (and remove the current files in ./updates/patcher)

Once the option has been selected, I'll need to update the buildbotcustom code.

After speaking with Callek over irc, the best bet are both options with the following changes: 1) a new repo is created under hg.mozilla.org/SeaMonkey called tools. i.e. hg.mozilla.org/SeaMonkey/tools 2) Callek's tools repo is migrated/moved to it. 3) the patcher files and patcher configs are moved to a directory under the tools/ repo. (Where? the option#2 patch above puts them in release/updates, but there are already files in there (patcher-config files...*, so I'm going to reorganize the release/updates folder.) * - apparently (from what I remember), we actually update both the CVS patcher-config files and Callek's tools/release/updates/ patcher-config files. (which is why when something chokes, we need to back out both CVS and Callek's Tools changes) I think (and I'll need Callek's confirmation) that we can even do away with the cvs patcher and have whatever reading off the CVS patcher-config changes to use Callek's tools patcher-config changes.. ) (or rather, the new tools/ repo.) any opinions/questions?

Additional thing I noticed. The patcher files refer to CVS and therefore, they must be modified to refer to mercurial.

This patch does the following: 1) Adds BootStrap, MozBuild, patcher folders to release/updates 2) Moves the original patcher-config files in releases/updates to release/updates/patcher-configs

Comment on attachment 8649128 [details] [diff] [review] buildbotcustom code changes (v1) This is a WIP: I had replaced the CVS checkout steps with copying the patcher/MozBuild/Bootstrap/patcher-configs to the build folder; but now, I realize that they must be within their original positions as they need to be diffed (particularly the patcher-configs) and the differences pushed to the repo.

This patch leaves all those patcher/MozBuild/Bootstrap/patcher-configs where they are and basically run the whole patcher process in tools/release/updates.

(In reply to Edmund Wong (:ewong) from comment #18) > Created attachment 8649159 [details] [diff] [review] > buildbotcustom code changes (v2) > > This patch leaves all those patcher/MozBuild/Bootstrap/patcher-configs where > they are and basically run the whole patcher process in > tools/release/updates. this is said being ignorant of what goes on in the patcher scripts (in PERL, no less), but a quick scan of patcher2.pl, it doesn't look like it uses the build/ stuff.. but not having touched perl that much, I'll need to really look at those scripts. (/me wonders if we can make those into python scripts?)

Needed to fix the path to the required sh files from within verify.sh

Comment on attachment 8650465 [details] [diff] [review] tools patch (v2) I bungled this patch. Need to redo this patch with the proper patcher-configs files.

Comment on attachment 8650501 [details] [diff] [review] tools patch (v3) Review of attachment 8650501 [details] [diff] [review]: ----------------------------------------------------------------- r+ if you do the following before checkin: cd <tools-repo-directory> find release/updates/patcher-configs/ ! -name *sea* # Confirm file list does not include any SeaMonkey patcher configs find release/updates/patcher-configs/ ! -name *sea* -delete # Update patch with this result

Comment on attachment 8650503 [details] [diff] [review] buildbotcustom code changes (v3) Review of attachment 8650503 [details] [diff] [review]: ----------------------------------------------------------------- I *think* this is all good, lets cross our fingers ::: process/factory.py @@ +5400,5 @@ > )) > self.addStep(TinderboxMockCommand( > name='diff_patcher_config', > command=['bash', '-c', > + '(hg diff "%s") && (grep \ instead of dropping -u use -U8 @@ +5506,5 @@ > env={'HGROOT': self.mozRepository, > 'PATH': '/usr/bin:/bin:/usr/sbin:/sbin:/usr/local/bin'}, > haltOnFailure=True, > timeout=3600, > + workdir='tools/release/updates/patcher', 'tools/release/updates' not +/patcher @@ +5593,5 @@ > name='patcher_download_builds', > command=command, > description=['patcher:', 'download builds'], > haltOnFailure=True, > + workdir='tools/release/updates/patcher', 'tools/release/updates' not +/patcher @@ +5611,5 @@ > name='patcher_create_patches', > command=command, > description=['patcher:', 'create patches'], > haltOnFailure=True, > + workdir='tools/release/updates/patcher', 'tools/release/updates' not +/patcher

Comment on attachment 8649129 [details] [diff] [review] buildbot-config changes (v1) Review of attachment 8649129 [details] [diff] [review]: ----------------------------------------------------------------- r- for the two mentioned issues. ::: seamonkey/config.py @@ +96,5 @@ > 'stage_server': 'stage.mozilla.org', > 'aus2_host': 'aus2-community.mozilla.org', > 'download_base_url': 'http://ftp.mozilla.org/pub/mozilla.org/seamonkey', > 'graph_server': 'graphs.mozilla.org', > + 'build_tools_repo_path': 'SeaMonkey/tools', don't change this *yet* ::: seamonkey/release_master.py @@ +244,5 @@ > buildNumber=releaseConfig['buildNumber'], > mozRepoPath=releaseConfig['mozillaRepoPath'], > inspectorRepoPath=releaseConfig['inspectorRepoPath'], > venkmanRepoPath=releaseConfig['venkmanRepoPath'], > chatzillaRepoPath=releaseConfig['chatzillaRepoPath'], there are more cases of cvsroot in this file you need to cull

Comment on attachment 8655472 [details] [diff] [review] [configs] buildbot-configs changes (v2) [checked-in] Review of attachment 8655472 [details] [diff] [review]: ----------------------------------------------------------------- Lets give this a go!

Comment on attachment 8655472 [details] [diff] [review] [configs] buildbot-configs changes (v2) [checked-in] Pushed to buildbot-configs: https://hg.mozilla.org/build/buildbot-configs/rev/5dc684b8f949

Comment on attachment 8655468 [details] [diff] [review] [custom] buildbotcustom code changes (v4) [checked-in] Pushed to buildbotcustom: https://hg.mozilla.org/build/buildbotcustom/rev/f6f4e701119e

Comment on attachment 8655449 [details] [diff] [review] [tools] patch (v4) [checked-in] Pushed to users/Callek_gmail.com/tools: https://hg.mozilla.org/users/Callek_gmail.com/tools/rev/4ae5a2886153

OTR, looking through the patches again, I realized last night that it's not enough. factory.py needs some path reference changes (patchConfigFile != patch- configs). patchConfigFile refers to the file which was previously in tools/release/updates/mozRelease*-<platform>.cfg. That's only just one thing. So back to the drawing board...

(In reply to Edmund Wong (:ewong) from comment #34) > OTR, looking through the patches again, I realized last night that it's not > enough. factory.py needs some path reference changes (patchConfigFile != > patch- > configs). patchConfigFile refers to the file which was previously > in tools/release/updates/mozRelease*-<platform>.cfg. > correction: tools/release/updates/mozRelease-seamonkey-branch-patcher2.cfg

(for post-land review)

Comment on attachment 8655813 [details] [diff] [review] [tools] correct path to retry.py command [checked-in] Pushed to tools: https://hg.mozilla.org/users/Callek_gmail.com/tools/rev/53913d27c44c

Comment on attachment 8656001 [details] [diff] [review] [tools] common/unpack.sh's mentions of common/ scripts needs changing. Pushed to tools: https://hg.mozilla.org/users/Callek_gmail.com/tools/rev/57632c74b44e

Comment on attachment 8656001 [details] [diff] [review] [tools] common/unpack.sh's mentions of common/ scripts needs changing. Backed out for a better solution: https://hg.mozilla.org/users/Callek_gmail.com/tools/rev/297721d736e4

Comment on attachment 8656032 [details] [diff] [review] [tools] verify.sh's references to common/ paths needs changing. (*) Pushed to tools: https://hg.mozilla.org/users/Callek_gmail.com/tools/rev/6353d7525521

(In reply to Edmund Wong (:ewong) from comment #45) > Comment on attachment 8656032 [details] [diff] [review] > [tools] verify.sh's references to common/ paths needs changing. (*) > > Pushed to tools: > https://hg.mozilla.org/users/Callek_gmail.com/tools/rev/6353d7525521 Backed this out. Pushed replacing patch: https://hg.mozilla.org/users/Callek_gmail.com/tools/rev/59ef785b7317 https://hg.mozilla.org/users/Callek_gmail.com/tools/rev/306ffea7d128

Comment on attachment 8656333 [details] [diff] [review] [tools] fix for final verification step (**) [checked-in] Pushed to tools: https://hg.mozilla.org/users/Callek_gmail.com/tools/rev/618386cc6361

Comment on attachment 8659782 [details] [diff] [review] [custom] Add -I to patcher command to get it to recognize the patcher files. [checked-in] Pushed to buildbotcustom: https://hg.mozilla.org/build/buildbotcustom/rev/97be4923e7fb

Comment on attachment 8655778 [details] [diff] [review] [custom] Use right user for commit during commit_patcher_config [checked-in] Pushed to buildbotcustom: https://hg.mozilla.org/build/buildbotcustom/rev/935f9d8784db

Comment on attachment 8659789 [details] [diff] [review] [custom] buildbotcustom code changes to fix verification step. (in conjunction with patch (*)) [checked-in] Pushed to buildbotcustom: https://hg.mozilla.org/build/buildbotcustom/rev/81b1838ceb86

Comment on attachment 8656332 [details] [diff] [review] [custom] buildbotcustom code changes to fix final verification step (used with **) This has been merged with this push: https://bugzilla.mozilla.org/attachment.cgi?id=8659789

Comment on attachment 8656341 [details] [diff] [review] [custom] buildbotcustom code changes to fix final verification step (v2) This has been merged into: https://bugzilla.mozilla.org/attachment.cgi?id=8659789

Closing this as resolved, its been for some time! At this point I dont reasonably feel I will be reviewing these patches, either way we are off from CVS \o/

Comment on attachment 8655778 [details] [diff] [review] [custom] Use right user for commit during commit_patcher_config [checked-in] Clearing review since this bug is now done (and has been for some time)

Comment on attachment 8655813 [details] [diff] [review] [tools] correct path to retry.py command [checked-in] Clearing review since this bug is now done (and has been for some time)

Comment on attachment 8656043 [details] [diff] [review] [tools] verify.sh's references to common/ paths needs changing. (*) [checked-in] Clearing review since this bug is now done (and has been for some time)

Comment on attachment 8656333 [details] [diff] [review] [tools] fix for final verification step (**) [checked-in] Clearing review since this bug is now done (and has been for some time)

Comment on attachment 8659120 [details] [diff] [review] remove cvs, rsh package from mock package list (v1) Clearing review since this bug is now done (and has been for some time)

Comment on attachment 8659789 [details] [diff] [review] [custom] buildbotcustom code changes to fix verification step. (in conjunction with patch (*)) [checked-in] Clearing review since this bug is now done (and has been for some time)