Comments on End Point Blog: PostgreSQL switches to Git

Nicolas: Not in the Postgres community it seems. I...

2010-05-24T09:17:06.875-04:00

Nicolas: Not in the Postgres community it seems. I've never heard anyone talk about it, and a mailing list search (http://postgresql.markmail.org/search/?q=gerrit) finds no relevant hits.

Anyone used Gerrit for code reviews?

2010-05-20T14:18:47.047-04:00

Anyone used Gerrit for code reviews?

Greg, yes, the URLs are nice, but in my experience...

2010-05-20T11:59:20.117-04:00

Greg, yes, the URLs are nice, but in my experience code review happens much more when the diff is right there inline (not in an attachment, right inline). Even "huge" diffs are rarely all that huge, especially compared to some of the bloated HTML email kicking around the net these days. :)

In any case, it doesn't need to be either/or. It'd be easy to have one list that gets commit info + link only, and another that also gets inline diffs, so people can choose which they prefer.

The move to Git is great news, by the way!

At a minimum we can provide a gitweb URL, so that ...

2010-05-20T01:42:32.339-04:00

At a minimum we can provide a gitweb URL, so that it's as simple as clicking on the link in the e-mail if you actually want to review the code. That doesn't risk sending out giant diffs and correspondingly bloating the mailing list, while still making it easy to see them if you want to.

@David: the swap from the current incrementally updated repo to a better quality one-time conversion is only going to happen at the switchover. So there will be a single bit of disruption that changes all the commit history, but you'll know it's coming long in advance, and the incremental one will be available until then.

The mailing list for commits with inline diffs wou...

2010-05-19T18:36:05.039-04:00

The mailing list for commits with inline diffs would be easy enough to create on our own. I'm not sure what good commit notifications without diffs are, really. They either mean I have to manually go look up the diff, or I just don't and then there's less code review going on.

The hashes on git.postgresql.org will change. No w...

2010-05-19T17:21:42.582-04:00

The hashes on git.postgresql.org will change. No way around that, but it's a very useful one-time pain!

So are they rebuilding the master repository, comp...

2010-05-19T17:16:59.949-04:00

So are they rebuilding the master repository, compared to the version they've had on git.postgresql.org, or keeping the incremental CVS import until switchover so $hackers don't have to be sad their SHA1s are all changing?