Patchbot ticket status

[Keshav Kini]

Hi,

I've been playing around with the installable patchbot code Robert put
up at http://trac.sagemath.org/sage_trac/ticket/12486 . Something I'm a
bit confused about is how the patchbot decides what color circle to
display on a trac ticket page, now that there are multiple patchbot
clients rather than a single one.

Take for example #11310. I just forcibly ran my patchbot on this ticket.
It's never been set to needs_review before so I guess it never got
tested by any other patchbots automatically. Thus my patchbot's result
(ApplyFailed) is the only patchbot result known for that ticket. Yet the
ticket's status remains "gray circle", which I guess means "not yet
tested". Why? Is it because I'm running a development version and not a
stable version?

It seems to me like the most natural way to determine the overall status
of the ticket is to find the latest version of Sage on which the ticket
has been patchbot-tested, and then determine the overall status from the
results of tests on that version of Sage only. In fact, I thought that's
exactly what the patchbot does, but this ticket seems to demonstrate
otherwise.

The pages in question:

http://trac.sagemath.org/sage_trac/ticket/11310
http://patchbot.sagemath.org/ticket/11310/

I figured I'd ask here since there are not too many people CC'd to
#12486 and there is probably some amount of patchbot-related lore
floating around sage-devel...

-Keshav

----
Join us in #sagemath on irc.freenode.net !

[David Loeffler]

The grey circle is because the patches now on the ticket aren't the
same ones that patchbot ran its tests with (note the word "mismatch"
in red on the patchbot page).

David

> --
> To post to this group, send an email to sage-...@googlegroups.com
> To unsubscribe from this group, send an email to sage-devel+...@googlegroups.com
> For more options, visit this group at http://groups.google.com/group/sage-devel
> URL: http://www.sagemath.org

[Keshav Kini]

David Loeffler <d.a.lo...@warwick.ac.uk> writes:
> The grey circle is because the patches now on the ticket aren't the
> same ones that patchbot ran its tests with (note the word "mismatch"
> in red on the patchbot page).

Unfortunately it looks like the patch was updated a mere 11 minutes
after I posted this thread, so you can't see exactly what I saw. The
circle was grey before the patch was updated, and "(mismatch)" did not
appear on the patchbot's page for the ticket.

[Robert Bradshaw]

On Tue, May 15, 2012 at 2:33 AM, Keshav Kini <kesha...@gmail.com> wrote:
>
> Hi,
>
> I've been playing around with the installable patchbot code Robert put
> up at http://trac.sagemath.org/sage_trac/ticket/12486 . Something I'm a
> bit confused about is how the patchbot decides what color circle to
> display on a trac ticket page, now that there are multiple patchbot
> clients rather than a single one.
>
> Take for example #11310. I just forcibly ran my patchbot on this ticket.
> It's never been set to needs_review before so I guess it never got
> tested by any other patchbots automatically. Thus my patchbot's result
> (ApplyFailed) is the only patchbot result known for that ticket. Yet the
> ticket's status remains "gray circle", which I guess means "not yet
> tested". Why? Is it because I'm running a development version and not a
> stable version?

Yep, it's hard coded at the moment.

> It seems to me like the most natural way to determine the overall status
> of the ticket is to find the latest version of Sage on which the ticket
> has been patchbot-tested, and then determine the overall status from the
> results of tests on that version of Sage only. In fact, I thought that's
> exactly what the patchbot does, but this ticket seems to demonstrate
> otherwise.

I've been meaning to change it to do just this.

> The pages in question:
>
> http://trac.sagemath.org/sage_trac/ticket/11310
> http://patchbot.sagemath.org/ticket/11310/
>
> I figured I'd ask here since there are not too many people CC'd to
> #12486 and there is probably some amount of patchbot-related lore
> floating around sage-devel...
>
> -Keshav
>
> ----
> Join us in #sagemath on irc.freenode.net !
>

[Keshav Kini]

Robert Bradshaw <robe...@math.washington.edu> writes:
> On Tue, May 15, 2012 at 2:33 AM, Keshav Kini <kesha...@gmail.com> wrote:
>> Take for example #11310. I just forcibly ran my patchbot on this ticket.
>> It's never been set to needs_review before so I guess it never got
>> tested by any other patchbots automatically. Thus my patchbot's result
>> (ApplyFailed) is the only patchbot result known for that ticket. Yet the
>> ticket's status remains "gray circle", which I guess means "not yet
>> tested". Why? Is it because I'm running a development version and not a
>> stable version?
>
> Yep, it's hard coded at the moment.

Aha, I see. Thanks!

>> It seems to me like the most natural way to determine the overall status
>> of the ticket is to find the latest version of Sage on which the ticket
>> has been patchbot-tested, and then determine the overall status from the
>> results of tests on that version of Sage only. In fact, I thought that's
>> exactly what the patchbot does, but this ticket seems to demonstrate
>> otherwise.
>
> I've been meaning to change it to do just this.

Well, for now could you quickly change the hard-coded stable version to
5.0? :) It happens to be the same as the latest development version at
the moment, so at least the behavior should then be correct for a little
while...