On Tue, 4 Mar 2025 at 18:25, Peter Geoghegan <pg@bowt.ie> wrote:
> From the looks of the screen shot that you posted (can't seem to find
> the same dashboard view on https://commitfest-test.postgresql.org?),
The dashboard is only available if you login. You probably have to
create an account to do so, because the staging auth and prod auth
systems are separate. Then you can mark yourself as author/reviewer of
a few patches to see what it would look like.
> this is *exactly* what I had in mind -- I don't know what I said that
> you haven't fully taken into account here? It's just a screen shot,
> but as far as it goes it looks great.
I think I grouped & ordered things slightly different than you
described. There are 4 groups (all of which are in the screenshot).
And then within each group patches are ordered like this:
1. Lowest max of "failing since", "needs rebase since", "time since
its commitfest was closed" at the top. NULLs (i.e. healthy patches)
are first.
2. If 1 tied (usually nulls) ordered by their commitfest startdate
(most recent startdate first)
3. If 2 ties, then patches are ordered by "most recent email"
So patches with failing CI in the "in progress cf" will sort below
healthy patches in the "open cf". I don't think you necessarily said
that, but this seemed nice to me. And it's easy to spot which patches
are for which CF because of the color coded CF labels.
This kind of sorting is possibly worth tweaking a bit after people
start using this and running into annoyances or unexpected sorts in
practice. Some other thing that's missing is the ability to "filter by
commitfest".
> Did you mean that you have general doubts about the general quality of
> the dashboard code? As in, the code itself is rough?
Nah, that's not what I meant.
> It'd be nice if at some point you also added the ability to
> star/favorite/like patches -- I'm thinking of something that worked a
> little bit like starring a gmail thread. Any such patches would appear
> towards the end of the dashboard page, in its own section,
> independently of whether I as a user am involved or not involved in
> the patch. This would be private information, visible only to the
> individual user that favorited the patch -- a mere bookmark.
Yeah, I had similar ideas.
