Re: Automated build-and-test summary report (2007/11/05)

Re: Automated build-and-test summary report (2007/11/05)

[Mark Wielaard]

[-- Attachment #1: Type: text/plain, Size: 929 bytes --]

Hi Kris,

On Mon, 2007-11-05 at 09:00 -0500, Kris Van Hees wrote:
> More results (and detailed reports) can be found on the labrat website:
> 
>     http://build.alchar.org/
> 
> ------------------------------------------------------------------------------
> Package:        frysk_fresh
> Previous build: 2007/11/04 00:24:27 - 2007/11/04 00:26:26
> Current build:  2007/11/05 00:24:22 - 2007/11/05 00:26:10
> Host:           ca-tools2
> Platform:       Linux 2.6.22.9-61.fc6 i386 (FC6)
> URL:            http://build.alchar.org/cgi/showBuild?host=ca-tools2&pkg=frysk_fresh&date=20071105-002422
> First failure:  no changes - not built
> Test result:    0: 0 +, 0 -, 0 S, 0 E
> 
>   Old   -> New   Test
>   -----    ----- ----------------------------------
>   (Previous build did not run tests.)

The logs are empty (or actually not there on the page above).
Do you know what happened?

Thanks,

Mark

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: Automated build-and-test summary report (2007/11/05)

[Kris Van Hees]

Well, the build report is definitely at the mentioned URL (just verified that)
although it might seem like it is missing information.  I.e. the build status
ends with the checksum phase.  That is a typical situation where no changes
were found between the previous source tree and the current one.  If it seems
confusing, I'll make changes to the reporting to make that more clear.

The fact that the (usually) downloadable logs are not being shown as links
(that section is empty) is a bug.  I'll send notice when it is resolved.

	Cheers,
	Kris

On Tue, Nov 06, 2007 at 11:25:29AM +0100, Mark Wielaard wrote:
> Hi Kris,
> 
> On Mon, 2007-11-05 at 09:00 -0500, Kris Van Hees wrote:
> > More results (and detailed reports) can be found on the labrat website:
> > 
> >     http://build.alchar.org/
> > 
> > ------------------------------------------------------------------------------
> > Package:        frysk_fresh
> > Previous build: 2007/11/04 00:24:27 - 2007/11/04 00:26:26
> > Current build:  2007/11/05 00:24:22 - 2007/11/05 00:26:10
> > Host:           ca-tools2
> > Platform:       Linux 2.6.22.9-61.fc6 i386 (FC6)
> > URL:            http://build.alchar.org/cgi/showBuild?host=ca-tools2&pkg=frysk_fresh&date=20071105-002422
> > First failure:  no changes - not built
> > Test result:    0: 0 +, 0 -, 0 S, 0 E
> > 
> >   Old   -> New   Test
> >   -----    ----- ----------------------------------
> >   (Previous build did not run tests.)
> 
> The logs are empty (or actually not there on the page above).
> Do you know what happened?
> 
> Thanks,
> 
> Mark

Re: Automated build-and-test summary report (2007/11/05)

[Mark Wielaard]

[-- Attachment #1: Type: text/plain, Size: 894 bytes --]

Hi Kris,

On Tue, 2007-11-06 at 06:41 -0500, Kris Van Hees wrote:
> Well, the build report is definitely at the mentioned URL (just verified that)
> although it might seem like it is missing information.  I.e. the build status
> ends with the checksum phase.  That is a typical situation where no changes
> were found between the previous source tree and the current one.  If it seems
> confusing, I'll make changes to the reporting to make that more clear.

OK, so "First failure:  no changes - not built" actually means that the
sources didn't change from the last time the builder ran and so it
wasn't build. Correct? If so the "First Failure" part confused me.

> The fact that the (usually) downloadable logs are not being shown as links
> (that section is empty) is a bug.  I'll send notice when it is resolved.

OK, yes, that was the other confusion.

Thanks,

Mark

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: Automated build-and-test summary report (2007/11/05)

[Kris Van Hees]

On Tue, Nov 06, 2007 at 01:22:56PM +0100, Mark Wielaard wrote:
> OK, so "First failure:  no changes - not built" actually means that the
> sources didn't change from the last time the builder ran and so it
> wasn't build. Correct? If so the "First Failure" part confused me.

Yes, in the email report the first failure entry actually gives the first
stage in the build-and-run instance where a failure was detected.  There are
actually three possible entry types here (both for the current result and for
the previous result):

	- no failures: this obviously is a rare entry, because it means that
		all stages completed without any problem
	- no changes - not built: this is reported when there were not changes
		found in the source tree during the build-and-test run; labrat
		doesn't perform a build/test run unless at least something in
		the source tree changed
	- a stage name: the most common case, reporting the first stage that
		failed

I guess it can be a bit confusing in cases like:

	First failure:  test (was no changes - not built)

especially because it might (incorrectly0 read as if the (was...) part is
actually an explanation of why 'test' was the first failure, rather than
reflecting the previous state (result of the previous build we're comparing
to).

Would it be more clear if I changed that to be:

	First failure:  test (previously: no changes - not built)

> > The fact that the (usually) downloadable logs are not being shown as links
> > (that section is empty) is a bug.  I'll send notice when it is resolved.
> 
> OK, yes, that was the other confusion.

Being worked on right now.  The nice part of task separation is that one thing
failing in the reporting engine won't stop the rest from work.  The less nice
part of task separation is that if one thing in the reporting engine fails and
the rest keeps working, it is not necessarily noticed (by me) that there is a
problem :)  Win a few, lose a few.

	Cheers,
	Kris

Re: Automated build-and-test summary report (2007/11/05)

[Mark Wielaard]

[-- Attachment #1: Type: text/plain, Size: 860 bytes --]

Hi Kris,

Thanks for that explanation. Much clearer now!

On Tue, 2007-11-06 at 08:44 -0500, Kris Van Hees wrote:
> Would it be more clear if I changed that to be:
> 
> 	First failure:  test (previously: no changes - not built)

Yes, adding previously is much better.
But maybe just put it on its own line?

  First failure this run     : test
  First failure previous run : no changes - not built

(It would be nice to have an entry also for the last time built and
where it failed that time.)

I think the real confusion comes from the "-".
Why not replace it with "so" (no changes so not built).
Or just spell it out completely "not built because no sources changed"
to be completely clear.

All the above is clearly nitpicking though. Now that I have read your
explanation I can easily parse any of the outputs.

Cheers,

Mark

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]