> The patch doesn't apply - I suppose the patch is relative a forked postgres
No, the authors just used a little outdated `master` branch. I successfully applied it against 31d8d474 and then rebased to the latest master (62ce0c75). The new version is attached.
Not 100% sure if my rebase is correct since I didn't invest too much time into reviewing the code. But at least it passes `make installcheck` locally. Let's see what cfbot will tell us.
Thank you very much! We'll do the same rebase and check this soon. Let's use v10 now.
> I encourage trying to break down the patch into smaller incrementally useful > pieces. E.g. making all the SLRUs 64bit would be a substantial and > independently committable piece.
Completely agree. And the changes like:
+#if 0 /* XXX remove unit tests */
... suggest that the patch is pretty raw in its current state.
Pavel, Maxim, don't you mind me splitting the patchset, or would you like to do it yourself and/or maybe include more changes? I don't know how actively you are working on this.
I don't mind and appreciate you joining this. If you split this we'll just make the next versions based on it. Of course, there is much to do and we work on this patch, including pg_upgrade test fail reported by Justin, which we haven't had time to concentrate on before. We try to do changes in small batches so I consider we can manage parallel changes. At least I read this thread very often and can answer soon, even if our new versions of patches are not ready.
Again I consider the work you propose useful and big thanks to you, Alexander!