Re: [Libreoffice-qa] Tinderbox status

2011-12-01 Thread Pedro Lino
Excellent! Thank you all for the answers ;)

Now onto some real bug swatting :)
___
List Name: Libreoffice-qa mailing list
Mail address: Libreoffice-qa@lists.freedesktop.org
Change settings: http://lists.freedesktop.org/mailman/listinfo/libreoffice-qa
Problems? http://www.libreoffice.org/get-help/mailing-lists/how-to-unsubscribe/
Posting guidelines + more: http://wiki.documentfoundation.org/Netiquette
List archive: http://lists.freedesktop.org/archives/libreoffice-qa/


Re: [Libreoffice-qa] Tinderbox status

2011-12-01 Thread Christian Lohmaier
Hi Pedro, *,

On Thu, Dec 1, 2011 at 1:56 PM, Pedro Lino  wrote:
> [...]
> It was my mistake. I mixed results, But you can see that for Win
> machine #6 the current error number is 375 and the previous successful
> build had 176
> http://tinderbox.libreoffice.org/cgi-bin/tinder.cgi?tree=MASTER&start-time=1322562647&display-hours=50

So then the already mentioned boost comes into play again. Boost
tarball is extracted, and the list of included files is printed to the
log.
The directory structure contains:
boost_1_44_0/libs/exception/ - and the parser triggers on "exception"
(and other terms). So just by extracting the boost tarball you get
lots and lots of "errors".

So while the increase in errors should alarm you to look at the logs,
after having that log you can lay back and don't need to worry.

Once the ErrorParser is taught to also ingore boost 1_44_0 and not
only older version of boost, the error-count will be reduced
significantly again.

ciao
Christian
___
List Name: Libreoffice-qa mailing list
Mail address: Libreoffice-qa@lists.freedesktop.org
Change settings: http://lists.freedesktop.org/mailman/listinfo/libreoffice-qa
Problems? http://www.libreoffice.org/get-help/mailing-lists/how-to-unsubscribe/
Posting guidelines + more: http://wiki.documentfoundation.org/Netiquette
List archive: http://lists.freedesktop.org/archives/libreoffice-qa/


Re: [Libreoffice-qa] Tinderbox status

2011-12-01 Thread Markus Mohrhard
2011/12/1 Pedro Lino :
>>> So if Tinderbox #9 finished successfully 2 days ago with 176 errors
>>> and today with 846(!!!), doesn't that sound suspicious?
>
>> Yes, it does - but only if it is the same machine, the same builder.
>> In your initial post you were comparing different builders, and there
>> it is rather irrelevant.
>
> No I wasn't. I was providing two examples of the same problem. On the
> MinGW errors increased from 16 to 220 and on the MSVC from 176 to 375
> (846 was copied from the wrong column by mistake. My bad :) ).
>
>> I don't see a previous successful entry for #9, so nothing to compare
>> with right now, and I surely don't remember the error count from
>> before the rename :-)
>
> It was my mistake. I mixed results, But you can see that for Win
> machine #6 the current error number is 375 and the previous successful
> build had 176
> http://tinderbox.libreoffice.org/cgi-bin/tinder.cgi?tree=MASTER&start-time=1322562647&display-hours=50
>

It was a build system change that switched to a more verbose output.
Nothing to worry about.

As long as a build is marked as succesful you don't need to worry
about anything. There might still be problems with it but then
normally you can't see them at the tinderbox page and only test if
they start correctly.

Markus
___
List Name: Libreoffice-qa mailing list
Mail address: Libreoffice-qa@lists.freedesktop.org
Change settings: http://lists.freedesktop.org/mailman/listinfo/libreoffice-qa
Problems? http://www.libreoffice.org/get-help/mailing-lists/how-to-unsubscribe/
Posting guidelines + more: http://wiki.documentfoundation.org/Netiquette
List archive: http://lists.freedesktop.org/archives/libreoffice-qa/


Re: [Libreoffice-qa] Tinderbox status

2011-12-01 Thread Pedro Lino
>> So if Tinderbox #9 finished successfully 2 days ago with 176 errors
>> and today with 846(!!!), doesn't that sound suspicious?

> Yes, it does - but only if it is the same machine, the same builder.
> In your initial post you were comparing different builders, and there
> it is rather irrelevant.

No I wasn't. I was providing two examples of the same problem. On the
MinGW errors increased from 16 to 220 and on the MSVC from 176 to 375
(846 was copied from the wrong column by mistake. My bad :) ).

> I don't see a previous successful entry for #9, so nothing to compare
> with right now, and I surely don't remember the error count from
> before the rename :-)

It was my mistake. I mixed results, But you can see that for Win
machine #6 the current error number is 375 and the previous successful
build had 176
http://tinderbox.libreoffice.org/cgi-bin/tinder.cgi?tree=MASTER&start-time=1322562647&display-hours=50

Cheers,
Pedro
___
List Name: Libreoffice-qa mailing list
Mail address: Libreoffice-qa@lists.freedesktop.org
Change settings: http://lists.freedesktop.org/mailman/listinfo/libreoffice-qa
Problems? http://www.libreoffice.org/get-help/mailing-lists/how-to-unsubscribe/
Posting guidelines + more: http://wiki.documentfoundation.org/Netiquette
List archive: http://lists.freedesktop.org/archives/libreoffice-qa/


Re: [Libreoffice-qa] Tinderbox status

2011-12-01 Thread Christian Lohmaier
Hi Pedro, *,

On Thu, Dec 1, 2011 at 1:25 PM, Pedro Lino  wrote:
>
>>> I.e. does the number of errors
>>> potentially affect the quality and reliability of the binaries?
>>
>> Not necessarily. But it would be suspicious if for example the Mac
>> ones that are below 10 "errors" suddenly spike to 50 or more and still
>> be green. Then it is worth to have a look what triggered the parser,
>> and either silence the parser or fix the code.
>
> So if Tinderbox #9 finished successfully 2 days ago with 176 errors
> and today with 846(!!!), doesn't that sound suspicious?

Yes, it does - but only if it is the same machine, the same builder.
In your initial post you were comparing different builders, and there
it is rather irrelevant.
When you want to compare different bots, you need to actually look at
the errors (view the brief/Summary log)
I don't see a previous successful entry for #9, so nothing to compare
with right now, and I surely don't remember the error count from
before the rename :-)

ciao
Christian
___
List Name: Libreoffice-qa mailing list
Mail address: Libreoffice-qa@lists.freedesktop.org
Change settings: http://lists.freedesktop.org/mailman/listinfo/libreoffice-qa
Problems? http://www.libreoffice.org/get-help/mailing-lists/how-to-unsubscribe/
Posting guidelines + more: http://wiki.documentfoundation.org/Netiquette
List archive: http://lists.freedesktop.org/archives/libreoffice-qa/


Re: [Libreoffice-qa] Tinderbox status

2011-12-01 Thread Pedro Lino
Hi Christian

>> I.e. does the number of errors
>> potentially affect the quality and reliability of the binaries?
>
> Not necessarily. But it would be suspicious if for example the Mac
> ones that are below 10 "errors" suddenly spike to 50 or more and still
> be green. Then it is worth to have a look what triggered the parser,
> and either silence the parser or fix the code.

So if Tinderbox #9 finished successfully 2 days ago with 176 errors
and today with 846(!!!), doesn't that sound suspicious?

Cheers,
Pedro
___
List Name: Libreoffice-qa mailing list
Mail address: Libreoffice-qa@lists.freedesktop.org
Change settings: http://lists.freedesktop.org/mailman/listinfo/libreoffice-qa
Problems? http://www.libreoffice.org/get-help/mailing-lists/how-to-unsubscribe/
Posting guidelines + more: http://wiki.documentfoundation.org/Netiquette
List archive: http://lists.freedesktop.org/archives/libreoffice-qa/


Re: [Libreoffice-qa] Tinderbox status

2011-12-01 Thread Stephan Bergmann

On 12/01/2011 09:51 AM, Pedro Lino wrote:

One question: the MinGW finished with 220 errors while on 11-30 it
managed to build with 16 errors; the MSVC Win_9 machine finished
successfully with 846(!) errors but the previous successful build was
managed by Win_6 with "only" 176 errors.

Is it a regression, or it doesn't really matter the number of errors
as long as it is successful? I.e. does the number of errors
potentially affect the quality and reliability of the binaries?


Those "number of errors" are largely useless.  IIUC, the tinderbox 
software scans the build log for suspicious words and counts each 
occurrence of such a word as an error.  Like each compilation of a file 
in the boost/exception directory triggers this.


Stephan
___
List Name: Libreoffice-qa mailing list
Mail address: Libreoffice-qa@lists.freedesktop.org
Change settings: http://lists.freedesktop.org/mailman/listinfo/libreoffice-qa
Problems? http://www.libreoffice.org/get-help/mailing-lists/how-to-unsubscribe/
Posting guidelines + more: http://wiki.documentfoundation.org/Netiquette
List archive: http://lists.freedesktop.org/archives/libreoffice-qa/


[Libreoffice-qa] Tinderbox status

2011-12-01 Thread Pedro Lino
Hi guys

I'm glad to see that the Win Tinderboxes are finishing successfully again.

One question: the MinGW finished with 220 errors while on 11-30 it
managed to build with 16 errors; the MSVC Win_9 machine finished
successfully with 846(!) errors but the previous successful build was
managed by Win_6 with "only" 176 errors.

Is it a regression, or it doesn't really matter the number of errors
as long as it is successful? I.e. does the number of errors
potentially affect the quality and reliability of the binaries?

Regards,
Pedro
___
List Name: Libreoffice-qa mailing list
Mail address: Libreoffice-qa@lists.freedesktop.org
Change settings: http://lists.freedesktop.org/mailman/listinfo/libreoffice-qa
Problems? http://www.libreoffice.org/get-help/mailing-lists/how-to-unsubscribe/
Posting guidelines + more: http://wiki.documentfoundation.org/Netiquette
List archive: http://lists.freedesktop.org/archives/libreoffice-qa/