CI as debugging aid?

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
4 messages Options
Reply | Threaded
Open this post in threaded view
|

CI as debugging aid?

Frank Shearar-3
It just occurred to me that we could possibly use CI jobs as a means
of debugging naughty plugins.

Recall that we had LargePositiveInteger test failures crashing the VM
(http://squeakci.org/job/SqueakTrunk/67/console for instance)?

If we ran a full test suite on a VM with debugging support, we could
gain some insight into this particular, intermittent, problem.

Once 4.4's out the door, we'd just need to get our hands on a
debugging VM. If Eliot built one as part of his usual publishing
routine, we could just use that... *cough*.

frank

Reply | Threaded
Open this post in threaded view
|

Re: CI as debugging aid?

David T. Lewis
On Wed, Dec 19, 2012 at 01:21:31PM +0000, Frank Shearar wrote:

> It just occurred to me that we could possibly use CI jobs as a means
> of debugging naughty plugins.
>
> Recall that we had LargePositiveInteger test failures crashing the VM
> (http://squeakci.org/job/SqueakTrunk/67/console for instance)?
>
> If we ran a full test suite on a VM with debugging support, we could
> gain some insight into this particular, intermittent, problem.
>
> Once 4.4's out the door, we'd just need to get our hands on a
> debugging VM. If Eliot built one as part of his usual publishing
> routine, we could just use that... *cough*.

If you are thinking of the assert-enabled Cog that Eliot describes here:

http://lists.squeakfoundation.org/pipermail/squeak-dev/2012-December/166750.html 
 
You can probably just use the one that Eliot provided.

I expect that this is just a matter of setting a couple of flags in the
configuration step of building Cog, so you (we) can probably do this
easily on the squeakci.org server to make your own debugging VM. We'll
need to confirm which flags to set of course.

I'm not sure how useful this would be, but it can't hurt to try it.

Next April 1, I think I'll release my SUnit enhancements to provide
support for "self shouldnt: #crash" in unit tests ;-)

Dave


Reply | Threaded
Open this post in threaded view
|

Re: CI as debugging aid?

Frank Shearar-3
ONE DAY, I will be able to reflexively hit the "reply all" button when
it needs hitting.

frank

---------- Forwarded message ----------
From: Frank Shearar <[hidden email]>
Date: 19 December 2012 18:05
Subject: Re: CI as debugging aid?
To: "David T. Lewis" <[hidden email]>


On 19 December 2012 17:56, David T. Lewis <[hidden email]> wrote:

> On Wed, Dec 19, 2012 at 01:21:31PM +0000, Frank Shearar wrote:
>> It just occurred to me that we could possibly use CI jobs as a means
>> of debugging naughty plugins.
>>
>> Recall that we had LargePositiveInteger test failures crashing the VM
>> (http://squeakci.org/job/SqueakTrunk/67/console for instance)?
>>
>> If we ran a full test suite on a VM with debugging support, we could
>> gain some insight into this particular, intermittent, problem.
>>
>> Once 4.4's out the door, we'd just need to get our hands on a
>> debugging VM. If Eliot built one as part of his usual publishing
>> routine, we could just use that... *cough*.
>
> If you are thinking of the assert-enabled Cog that Eliot describes here:
>
> http://lists.squeakfoundation.org/pipermail/squeak-dev/2012-December/166750.html
>
> You can probably just use the one that Eliot provided.

Yes, those are the VMs I mean. I'd just like to see one with every release.

> I expect that this is just a matter of setting a couple of flags in the
> configuration step of building Cog, so you (we) can probably do this
> easily on the squeakci.org server to make your own debugging VM. We'll
> need to confirm which flags to set of course.

I think there was some noise around setting up builds to test the Cog
VMs anyway, on vm-dev, wasn't there?

> I'm not sure how useful this would be, but it can't hurt to try it.
>
> Next April 1, I think I'll release my SUnit enhancements to provide
> support for "self shouldnt: #crash" in unit tests ;-)

:)

frank

> Dave
>

Reply | Threaded
Open this post in threaded view
|

Re: CI as debugging aid?

David T. Lewis
On Wed, Dec 19, 2012 at 06:06:21PM +0000, Frank Shearar wrote:

> ONE DAY, I will be able to reflexively hit the "reply all" button when
> it needs hitting.
>
> frank
>
> ---------- Forwarded message ----------
> From: Frank Shearar <[hidden email]>
> Date: 19 December 2012 18:05
> Subject: Re: CI as debugging aid?
> To: "David T. Lewis" <[hidden email]>
>
>
> On 19 December 2012 17:56, David T. Lewis <[hidden email]> wrote:
> > On Wed, Dec 19, 2012 at 01:21:31PM +0000, Frank Shearar wrote:
> >> It just occurred to me that we could possibly use CI jobs as a means
> >> of debugging naughty plugins.
> >>
> >> Recall that we had LargePositiveInteger test failures crashing the VM
> >> (http://squeakci.org/job/SqueakTrunk/67/console for instance)?
> >>
> >> If we ran a full test suite on a VM with debugging support, we could
> >> gain some insight into this particular, intermittent, problem.
> >>
> >> Once 4.4's out the door, we'd just need to get our hands on a
> >> debugging VM. If Eliot built one as part of his usual publishing
> >> routine, we could just use that... *cough*.
> >
> > If you are thinking of the assert-enabled Cog that Eliot describes here:
> >
> > http://lists.squeakfoundation.org/pipermail/squeak-dev/2012-December/166750.html
> >
> > You can probably just use the one that Eliot provided.
>
> Yes, those are the VMs I mean. I'd just like to see one with every release.
>
> > I expect that this is just a matter of setting a couple of flags in the
> > configuration step of building Cog, so you (we) can probably do this
> > easily on the squeakci.org server to make your own debugging VM. We'll
> > need to confirm which flags to set of course.
>
> I think there was some noise around setting up builds to test the Cog
> VMs anyway, on vm-dev, wasn't there?

http://squeakci.org/job/CogVM/

This is not intended to test the Cog VMs, but it does do a VM build in
order to verify that the "bleeding edge" sources are coherent enough
to produce a VM. So if you need a locally-built VM on that squeakci
server with some specific configuration (such as debugging support),
you can easily tweak that test build to produce whatever you want.

But to be clear: These are *not* VMs that are intended for distribution,
and nobody should use them as a replacement for the official distributions
on Eliot's site. Likewise, there is an automated build of the standard
interpreter VM at http://squeakci.org/job/InterpreterVM/ that does
a VM build to check source integrity, and these are *not* a replacement
for Ian's official distributions at http://squeakvm.org/unix/.

Dave