Linked, abandoned branches are too prominent

[Jonathan Lange]

On bug 506127, I was just asked whether I intended to make progress with my linked branch. Thing is, I explicitly marked that branch as Abandoned quite some time ago in order to forestall such questions. Still, it was a natural question to ask, since the link is exactly as prominent and colorful as a branch with a merge proposal that's waiting for review.

I've attached a screenshot of the relevant section.

[Deryck Hodge (deryck)]

I think we should just not show links to abandoned branches, period. I think showing rejected status for non-abandoned branches is fine. Would you agree?

[Jonathan Lange (jml)]

On Fri, Apr 9, 2010 at 4:37 PM, Deryck Hodge <email address hidden> wrote:
> I think we should just not show links to abandoned branches, period.  I
> think showing rejected status for non-abandoned branches is fine.  Would
> you agree?

Yep.

[Deryck Hodge (deryck)]

Combing my frustration at little bugs like this going unfixed and a need to code some today, I spent my lunch hour fixing this bug.

[Deryck Hodge (deryck)]

abentley and maxb had some reservations on IRC, which I understand. But I don't have the desire to pursue this beyond this drive-by lunch time fix. I welcome better solutions from the code team, though, and leave the more involved fix to them. :-)

[Aaron Bentley (abentley)]

I think that hiding the branch is not the right approach. The fact that the original developer stopped working on it does not imply that the branch is useless, and we should err on the side of showing things that may be useful.

However, we should prominently display the abandoned status. We might also want to reduce the visibility of the merge proposal, or hide it entirely. It implies that the branch is in development, which is at odds with the branch's status. (We should still allow traversal to the merge proposal from the branch page itself.)

If there were no merge proposal, the Abandoned status would be prominent, so as a first cut, I suggest that we ignore the merge proposals of abandoned branches in this display.

[Martin Pool (mbp)]

On 10 April 2010 05:23, Aaron Bentley <email address hidden> wrote:
> I think that hiding the branch is not the right approach.  The fact that
> the original developer stopped working on it does not imply that the
> branch is useless, and we should err on the side of showing things that
> may be useful.

I agree.

>
> However, we should prominently display the abandoned status.  We might
> also want to reduce the visibility of the merge proposal, or hide it
> entirely.  It implies that the branch is in development, which is at
> odds with the branch's status.  (We should still allow traversal to the
> merge proposal from the branch page itself.)
>
> If there were no merge proposal, the Abandoned status would be
> prominent, so as a first cut, I suggest that we ignore the merge
> proposals of abandoned branches in this display.

To me this should somehow tie to the mp status too. If the branch was
proposed and rejected, as jml's was, then it should still be visible
on the bug but it perhaps less prominent than something that actually
has a fix and needs to be reviewed or merged.

Then the interesting thing is whether jml's still going to work on it
or not - this is perhaps best modelled by the bugtask assignment.

Then we might like to know whether the approach in this branch was
useful at all, or whether someone needs to basically start from
scratch. Perhaps abandoned comes in there? But perhaps that's just
the difference between needs fixing and rejected? (It's unclear
though.)

--
Martin <http://launchpad.net/~mbp/>

[Jonathan Lange (jml)]

I likewise agree with Aaron – good catch.

We have other options, such as greying out the linked branch.

Note that I rejected the MP myself. As far as I know, it's the best way of withdrawing a merge proposal.

Martin, you make good points. Re the last paragraph, if we link to the merge proposal, then that's probably the single best thing we can do to help people evaluate whether the approach in a branch was useful at all. Even if we had a flag for "useful approach", I would assume that those who went before got it wrong and look at the MP anyway.

[Martin Pool (mbp)]

On 12 April 2010 23:29, Jonathan Lange <email address hidden> wrote:
> I likewise agree with Aaron – good catch.
>
> We have other options, such as greying out the linked branch.
>
> Note that I rejected the MP myself. As far as I know, it's the best way
> of withdrawing a merge proposal.
>
> Martin, you make good points. Re the last paragraph, if we link to the
> merge proposal, then that's probably the single best thing we can do to
> help people evaluate whether the approach in a branch was useful at all.
> Even if we had a flag for "useful approach", I would assume that those
> who went before got it wrong and look at the MP anyway.

Right, after working through it, I think we have gone as far as we can
easily go with structured data. At this point people need to look at
the branch to see if it's useful, or just ask jml. Conversations
among people as happened here are not something we have to avoid at
all costs ;-)

--
Martin <http://launchpad.net/~mbp/>