This sounds good, but isn't the repo name redundant given it is implied by the email going to commits@l.a.o? On Feb 19, 2016 4:38 PM, "Chris Hostetter" <hossman_luc...@fucit.org> wrote:
> > : https://git-wip-us.apache.org/docs/switching-to-git.html seems to > : suggest there is per project flexibility. Branch not one of the > : (currently) available variables though, no? > : > : +1 for "the branch be included in the subject" > > Thanks for finding that link Christine, > > I pinged #infra on HipChat to try and find the actual code in question to > see how hard it would be to add "branch" based variables so I could > propose a patch to infra rather then just a general "can we do this?" type > request, but aparently that code is ASF specific and lives in a private > infra repo, so only infra members can read/write. Gavin said new subject > variables are usually not a big deal though. > > That said, before I request any changes, I want to make sure I'm > not wasting the time of any infra volunteers -- so I'd like to make sure > we have some concensus on what we'd ideally like... > > > : Perhaps the script could only include the last N elements of the name, > : so we get lucene-5438-nrt-replication or > : jira/lucene-5438-nrt-replication instead of the full branch name. Or > : maybe a regex could be used to target refs\/.*?\/ (or something more > : complex) for removal -- for some of the existing branch names, having > : the last three path elements would be good, but for others, one or two > : would be better. > > good point ... given that this is a general infra tool for all projects, > and currently the only per-project configuration is (aparently) what the > subject should be comprised of, i'm hesitent to try and request a lot of > custom regex rules, and/or making any general assumptions about only using > the last "N" elements of the name. > > (a common workflow i've seen is things > like refs/head/jira/solr-xyz for a shared collaboration on that feature, > while refs/head/hossman/jira/solr-xyz might be my proposed new direction > for the code to take -- we wouldn't want those to get confused.) > > That said, i think it would totally make sense to request that > "%(branch)s" should refering to the full branch path, and > "%(shortbranch)s" should be the result of regex stripping > "^refs\/(heads\/)?" from the full branch path. > > So "refs/heads/branch_7_5 => "branch_7_5" > > But "refs/tags/releases/lucene-solr/7.5.0" > => "tags/releases/lucene-solr/7.5.0" > > : There is normally a fairly limited amount of space for the subject in > : the list view of an email client, so it seems like a good idea to keep > : it short but relevant. > > Agreed -- so perhaps we should also request reducing some other > redundencies? (ie: "git commit") > > > what do folks think about requesting as our pattern... > > "git: %(repo_name)s:%(shortbranch)s: %(subject)s" > > > With some examples of what that would look like for a handful of commits > from the past month... > > > git: lucene-solr:branch_5x: fix test bug, using different randomness when > creating the two IWCs > > http://mail-archives.apache.org/mod_mbox/lucene-commits/201602.mbox/%3Ca9231a5cb3444a9ba70f1b67658d2844%40git.apache.org%3E > > [1/3] git: lucene-solr:lucene-6835: cut back to > Directory.deleteFile(String); disable 'could not removed segments_N so I > don't remove any other files it may reference' heroics > > http://mail-archives.apache.org/mod_mbox/lucene-commits/201602.mbox/%3C68acb868408348da8941e473725abda0%40git.apache.org%3E > > [1/2] git: lucene-solr:master: LUCENE-7002: Fixed MultiCollector to not > throw a NPE if setScorer is called after one of the sub collectors is done > collecting. > > http://mail-archives.apache.org/mod_mbox/lucene-commits/201602.mbox/%3ca53313b79d1b4286a655b03d2e2b2...@git.apache.org%3E > > [2/2] git: lucene-solr:branch_5x: LUCENE-7002: Fixed MultiCollector to not > throw a NPE if setScorer is called after one of the sub collectors is done > collecting. > > http://mail-archives.apache.org/mod_mbox/lucene-commits/201602.mbox/%3c40d0b62a4e2245ff85211c4fe4401...@git.apache.org%3E > > > ...note in particular those last two emails. As I understand it they > were two commits from the same "push", on diff branches (the master change > and the 5x backport) ... which is now more clear with the branch name in > the subject. > > Are folks in favor of requesting this from infra? > > > > -Hoss > http://www.lucidworks.com/ > > --------------------------------------------------------------------- > To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org > For additional commands, e-mail: dev-h...@lucene.apache.org > >
