Re: [Libreoffice-qa] Reminder: QA Meeting on Wednesday

[m.a.riosv]

Thanks for the information.

But I'm subscribed to the list time ago.

Miguel Ángel.

Hi All,
> BTW would be nice listen from the candidates their opinion about concrete
> matters like this one.

Should have suggested. If you're really interested in knowing candidates
views on specific items you may want to email [board-discuss] list 

Warmest Regards,
Joel





--
View this message in context: 
http://nabble.documentfoundation.org/Libreoffice-qa-Reminder-QA-Meeting-on-Wednesday-tp4167253p4167538.html
Sent from the QA mailing list archive at Nabble.com.
___
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] minutes of ESC call ...

[Michael Meeks]

On Thu, 2015-11-26 at 19:03 +0100, Markus Mohrhard wrote:
> please add a performance test that ensures that future refactorings
> won't regress here. In general any patch claiming performance
> improvements should be accompanied by a performance test to make sure

Great idea =) and easy to do too; Kendy ?

> Related to the code I wanted to mention that the getenv calls should
> be cached in a static variable as they might become expensive when

I was concerned by that too =) I believe the plan is to bin those; but
yes - good to check the perf tests for sure; and to have some nice
before/after step here.

ATB,

Michael.

-- 
 michael.me...@collabora.com  <><, Pseudo Engineer, itinerant idiot

___
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] Reminder: QA Meeting on Wednesday

[Alexander Thurgood]

Le 26/11/2015 17:00, Cor Nouws a écrit :

Hi Cor,


> In the early days, there was the meme that we needed good bug reports.
> Then that changed to good sample documents; tThen finding the first
> affected version; and now bibisecting.

You forgot the "please provide a full backtrace with symbols" (yes, I
have seen some of my backtraces declared obsolete because they lacked
symbols) ;-)

Not to mention that OS specific bugs, such as ones that are seen only on
OSX, for example, often seem to have no bibisectable repo (there doesn't
seem to be anything earlier than 4.1 branch off at

http://dev-downloads.libreoffice.org/MacOS_Bibibsect/

Virtualy all, if not all, of the current font weight support problems on
OSX seem to have appeared with the move to 4.1, so we would need a
bibisect repo from 3.5 to 4.1 to even start looking at these particular
regressions. CoreText work at the time is said to be the culprit (at
least, that is the comment I have seen from one dev in some of these
reports).

Similar problems with access to complete bibisect repos on Windows
exist, as far as my limited understanding of the problem goes.

How are OS-specific bugs treated if there is no bibisectable repo  ?
Do they just pile up and get forgotten in a corner ?


Alex

___
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] LibreOffice releases (was: Reminder: QA Meeting on Wednesday)

[Bjoern Michaelsen]

Hi,

On Wed, Nov 25, 2015 at 04:27:58PM -0700, Pedro wrote:
> But since you can't ask the developers to do such a boring job, even that
> won't solve the problem...

Thats not entirely true. Yes, just asking some random developer to fix some
random regression (likely by another developer) is a very bad idea as it
assumes guilt by association. Bibisecting a regression and asking the developer
who causes the regression to fix it ... is less of a problem. There still might
be valid reasons from this developer not being able to care for this particular
regression at this point in time, so dont assume this to be a strictly mechanic
thing.

Note however that regressions _are_ being watched at the ESC. And people are
also watching who caused them (which is only possible when they are
bibisected) and where -- at least I am doing this already for quite a while. Im
not calling names in public though, as that would likely be very
counterproductive. It would only be a last resort if things go really out of
hands.

So QA can help here by really bibisecting bugs: Bibisected bugs tell developers
where we cant be braver and we are going too fast and loose. Good data on
bibisected regressions allows developer to handle things properly. However,
organizing the response has to happen between the developers (e.g. on the ESC)
as it would be foolish to assume a strictly mechanical handling to be helpful
here.

Best,

Bjoern

P.S.: If you want to raise visibility to regressions in a helpful way, I
suggest you start with a _positive_ motivation for developers. E.g. QA giving
out a badge each month for the three developers with (most simple approach):

- a/ the most commits
- b/ without a regression known to be bibisected down to a commit of him/her

I assume something like that might create some good and _positive_ motivation
in the right direction ...
___
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] LibreOffice releases (was: Reminder: QA Meeting on Wednesday)

[Pedro]

Hi Bjoern


Bjoern Michaelsen wrote
> Note however that regressions _are_ being watched at the ESC. And people
> are
> also watching who caused them (which is only possible when they are
> bibisected) and where -- at least I am doing this already for quite a
> while. Im
> not calling names in public though, as that would likely be very
> counterproductive. It would only be a last resort if things go really out
> of
> hands.

I'm glad to read this. Obviously the goal is not to point fingers (in fact
that would be a really bad move)


Bjoern Michaelsen wrote
> So QA can help here by really bibisecting bugs: Bibisected bugs tell
> developers
> where we cant be braver and we are going too fast and loose. Good data on
> bibisected regressions allows developer to handle things properly.

I'm well aware of that. However bibisecting is not a trivial task (in terms
of setting up a system and in terms of time). It would indeed be good to
have more people doing this but it is not so easy to recruit people when the
project's visibility is low. I'm on the QA mailing list for Apache
OpenOffice and some days (not every day, of course) there are 3 or 4 people
volunteering to join QA...


Bjoern Michaelsen wrote
> P.S.: If you want to raise visibility to regressions in a helpful way, I
> suggest you start with a _positive_ motivation for developers. E.g. QA
> giving
> out a badge each month for the three developers with (most simple
> approach):
> 
> - a/ the most commits
> - b/ without a regression known to be bibisected down to a commit of
> him/her
> 
> I assume something like that might create some good and _positive_
> motivation
> in the right direction ...

This is a positive suggestion.
Thank you! 

Best regards,
Pedro



--
View this message in context: 
http://nabble.documentfoundation.org/Libreoffice-qa-Reminder-QA-Meeting-on-Wednesday-tp4167253p4167478.html
Sent from the QA mailing list archive at Nabble.com.
___
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] Reminder: QA Meeting on Wednesday

[Cor Nouws]

Joel Madero wrote on 25-11-15 16:22:

> Feel free to email the developer list and try to recruit developers to
> do what you're asking. No one is stopping you from doing this :) Doing
> it when many (most) of the regressions are not fully triaged might be a
> frustrating process for all involved.

There is a gradually growing number of regressions. Also bibisected
regressions. I know the ESC has attention for this, and that the board
recently approved projects/investments that are intended to help to
change this trend.
Pedro has a point that he doesn't like to accept the situation, and -
learning from his contributions - he is helping a lot with good work here.
Alas no ready to use solution from me today :)
It's on my mind, as some of you may know from some grumbling mails that
I send now and then. I expect to have some room next week to spend more
time on thinking for this. E.g. the suggestion for positive stimulation,
as mentioned also by Bjoern.

Pedro: can you please ping me in a week or so?

Thanks,
Cor


-- 
Cor Nouws
GPD key ID: 0xB13480A6 - 591A 30A7 36A0 CE3C 3D28  A038 E49D 7365 B134 80A6
- vrijwilliger http://nl.libreoffice.org
- volunteer http://www.libreoffice.org
- The Document Foundation Membership Committee Member
___
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] Reminder: QA Meeting on Wednesday

[Cor Nouws]

Hi Joel,

Joel Madero wrote on 26-11-15 02:31:
> If QA members just bisected each regression (fully bisect) and
> prioritized correctly I *honestly* believe that the regression count
> would fall. I don't understand why this point is being ignored as it's
> literally *completely* in our control.

Sorry to say that I'm afraid that I can't support you in that believe.
Not at the moment.
In the early days, there was the meme that we needed good bug reports.
Then that changed to good sample documents; tThen finding the first
affected version; and now bibisecting.
Don't get me wrong: I don't say that good QA is very important. But
currently I think that doing proper triage on all partly triaged and
unconfirmed bugs would only increase the number of unresolved regressions :)

Ciao,
Cor


-- 
Cor Nouws
GPD key ID: 0xB13480A6 - 591A 30A7 36A0 CE3C 3D28  A038 E49D 7365 B134 80A6
- vrijwilliger http://nl.libreoffice.org
- volunteer http://www.libreoffice.org
- The Document Foundation Membership Committee Member
___
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] minutes of ESC call ...

[Michael Meeks]

* Present:
+ Cloph, Sophie, Stephan, Jan-Marek, JanI, Michael M, Miklos, Andras,
  Caolan, Bjoern, Lionel, Bubli, Armin, Michael S, Kendy, Eike, Robinson, 
Thorsten
 
* Completed Action Items:
+ UserAgent - drop bundled-languages (Michael)
+ UserAgent - produce a patch for review (Michael)
+ disable openGL by default for 5.0.4 (Cloph)
+ run the sw renaming script just before branch-off (Robinson)
[ needed a fix to a test document, accidentally touched,
  kendy fixed for the future ]
+ improve new contributors Wiki, write blogpost, promote git-review etc. 
(JanI)
[ v. open-ended: in-progress ]
+ setup FOSDEM Hackfest wiki page with dates (Samuel stole this AI ;) )
[ https://wiki.documentfoundation.org/Hackfest/FOSDEM2016 ]
 
* Pending Action Items:
+ come up with a proposal for "abandoned" / "insufficient data"
  resolution in the bug-flow graph (Robinson)
[ lively discussion, following up, decision by the end of the week ]
+ book FOSDEM conf. room for board meeting (Florian)
[ perhaps after not before to avoid hackfest conflict ]
+ list of developers who havn't contributed in a while (JanI)
+ check mDNSResponder & move it to external/ (Michael S.)
[ hidden 5x directories deep in impress ]
+ double-check libmar/src, update/src, remove the duplicate, etc. (Kendy)
 
* Release Engineering update (Cloph)
+ 4.4.7 RC2
+ tagged, Linux + Mac built.
+ Windows delayed; since VM died; buying replacement H/W etc.
+ will start build today.
+ 5.0.4 status
+ RC1 / freeze - Monday
+ tagging due later today / tomorrow.
+ all fixes from gerrit are integrated
+ very last minute fixes today / tomorrow.
+ 5.1.0 branch / feature-freeze: this week.
+ branch created, and beta1 tag this week.
+ windows build breakage just before tagging.
+ a buildfix1 tag -> corresponding fixes.
+ re-created tar-balls from this.
+ Avoiding large-scale cosmetic cleanups
+ until when ? (Stephan)
+ consensus so-far; try to do just before branch-off (Miklos)
+ one idea mass-run through code-formatting tool (Bjoern)
+ start with config / code-style we want to use
+ by the time we do the next release: finally look at it.
   + 6 months to bike-shed.
+ afraid, can't agree on include order, not ideal (Miklos)
+ removing parameters, defaulted etc. (Stephan)
+ calming down for a couple of point-release to help bug-fixers 
(Michael)
+ makes sense (Bjoern)
+ concern, never a good time for doing these (Stephan)
+ concern - size, and non-functionality are deadly annoying when
  back-porting bug-fixes (Michael)
+ if a volunteer; doing rename member variables (Bjoern)
+ if 1st commit - makes sense perhaps.
+ agree with a quiet period (Eike)
+ will pile up changes in gerrit; but need reviewing.
+ piling makes them un-merge-able over time (Bjoern)
+ say - we discourage until .2 (as in the past) (Miklos)
+ not black / white - please try to avoid those.
=> avoid large non-functional cleanups until x.y.2
+ Late Features:
+ OpenGL - switch to simpler double-buffering.
+ Slide Transitions: artwork & polish
[ Jay and Andreas making new images - much appreciated ]
+ UserAgent display & tweakage
[ in modulo tweaking, cf. discussing a setting later ]
+ calc s/w interpreter-related patch
[ also got in before the freeze; an 18x speedup on CPU for
  some test sheets; currently guarded by a variable - and
  subsetting to just some sheets.
 + plan to test with crash-testing sheets & enable by 
default
  Would like to look at it (Eike)
 + dislike env.vars set - avoids the unit-tests
+ plan to remove it before ship (Kendy)
+ turn into normal config & default -> testable.
+ would like to turn it on now - default on for 
user-testing (Kendy)
+ optional B2 in the release-plan (Cloph)
+ can have an intermediate release.
 + goal of the code - vectors of doubles
+ formula results (Michael)
+ moves big chunk of ptr chasing & branching from 
inner loop.
]
+ Android & iOS Remote
+ delayed, due to B1 delay.
 
* Idle re-work (Michael)
+ graphical event processing issue
+ scheduler cleanup
+ true 'idle' 

Re: [Libreoffice-qa] Reminder: QA Meeting on Wednesday

[Joel Madero]

Hi All,
> There is a gradually growing number of regressions. Also bibisected
> regressions. I know the ESC has attention for this, and that the board
> recently approved projects/investments that are intended to help to
> change this trend.
> Pedro has a point that he doesn't like to accept the situation, and -
> learning from his contributions - he is helping a lot with good work here.
> Alas no ready to use solution from me today :)
> It's on my mind, as some of you may know from some grumbling mails that
> I send now and then. I expect to have some room next week to spend more
> time on thinking for this. E.g. the suggestion for positive stimulation,
> as mentioned also by Bjoern.
>
> Pedro: can you please ping me in a week or so?

I would like to take part also - I've thought about a few new ideas that
I think might work :) I won't be available until December 8. Almost any
time (absent really really late to really really early) would work
between December 8 - 10. Shall we aim for a phone call? Google hangout?
IRC Chat?

I'm happy Pedro brought this up as I think collectively we can come up
with a solution.


Best,
Joel

___
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] minutes of ESC call ...

[Markus Mohrhard]

Hey Michael, Keny, Eike,

On Thu, Nov 26, 2015 at 5:51 PM, Michael Meeks 
wrote:

>
> + calc s/w interpreter-related patch
> [ also got in before the freeze; an 18x speedup on CPU for
>   some test sheets; currently guarded by a variable - and
>   subsetting to just some sheets.
>  + plan to test with crash-testing sheets & enable by
> default
>   Would like to look at it (Eike)
>  + dislike env.vars set - avoids the unit-tests
> + plan to remove it before ship (Kendy)
> + turn into normal config & default ->
> testable.
> + would like to turn it on now - default on for
> user-testing (Kendy)
> + optional B2 in the release-plan (Cloph)
> + can have an intermediate release.
>  + goal of the code - vectors of doubles
> + formula results (Michael)
> + moves big chunk of ptr chasing & branching
> from inner loop.
> ]
>


please add a performance test that ensures that future refactorings won't
regress here. In general any patch claiming performance improvements should
be accompanied by a performance test to make sure that we can still
refactor the code without someone complaining that we introduced huge
performance regression.

Related to the code I wanted to mention that the getenv calls should be
cached in a static variable as they might become expensive when called
repeatedly and that the new virtual method calls are not free so I would
like to see some before and after numbers for both the supposed 18x speedup
case and our normal matrix handling code.

Regards,
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] Reminder: QA Meeting on Wednesday

[m.a.riosv]

El 26/11/15 a las 2:31, Joel Madero escribió:
> Hi There,
>
> On 11/25/2015 02:04 PM, m.a.riosv wrote:
>> BTW would be nice listen from the candidates their opinion about
>> concrete matters like this one.
>
> I would be surprised if there was any deviation for any of the
> candidates - we've discussed this at length on the Board. Hell I just
> brought it up today (saying it came up again in QA) and well, there was
> unanimous consensus and hopefully one of the developers will write a
> nice blog post about this issue to put it to rest once and for all.
> Telling others what to do is not how we work. We have a strong culture
> of "the doers decide" and the "doers" stick with what they "do" - thus
> again, I really do think there is a solution here.

"Telling others what to do", nothing is further from my mindset.

For your words seems everything is cooked, then why do the voting?,
expending resources and time.

I don't care on to have the reason, but reach some kind of progress to solve
a problem, problems don't disappear by saying not and looking for other
side, usually only get worse.

>
> If QA members just bisected each regression (fully bisect) and
> prioritized correctly I *honestly* believe that the regression count
> would fall. I don't understand why this point is being ignored as it's
> literally *completely* in our control.

01/08/15 - jmadero@...com
"I skimmed this and really...this isn't how the team works. We don't have
someone come and "insist" and "dictate" how we do our work. If you want then
the best thing for you to do is "lead by example" and then if it works,
others will follow."

Please take in account for yourself. I can't find in bugzilla how many
bisects you have done along the last year.

The concept is clear, but why is a QA job?, collaboration it's a matter for
two or more. I don't like the work flow, so wrongly or not I choose what to
do, hope it is a right for everybody?

>
> Anywho, I like the spirited conversation I just don't see any actionable
> item. I suggested that Pedro go track down developers to agree with him
> but he quickly dismissed that (because he couldn't convince me?) and
> asked Robinson or someone else to do the work of bringing up to the ESC.
> The ESC has also talked about this at length and rejected it (on several
> occasions). I've spoken with many of the developers who join the ESC and
> all of them reject this concept of QA dictating what they do, or that
> there should be any quasi-dictating by having "fix builds" or the like.
> But again, it's not outside of any of your powers to broach the subject
> yourselves on the dev list and pitch *concrete actionable ideas*.

I never talked about dictate no one what to do, who am I for that?, but who
choose what to do, must follow the project norms/rules or how you want call
it, in fact there are rules.
You can buy the car that you like or can, but you must follow the traffic
norms.

>
> At the same time, I highly encourage thinking about compromise positions
> - which I think I have actually given a good compromise. If we can go to
> the ESC and say "look our QA team is frantically doing *comprehensive
> triaging* and bisecting each and every regression, can you take over
> now? That seems like a more positive way to move things forward.

I have expressed some ideas several times, basically about if something can
be done to prevent the bugs, but you and by your comments everybody has only
an answer, all to the bin. So IMHO the ball is in your roof, but please
clean from your mind that QA people, at least me not, are here for your
, it's for the project.

An example, one dev implements a new enhancement, a report comes in
bugzilla, a QA person confirm and bisect, sometimes a lot of their free time
for that, the dev patch the issue, also sometimes, with a simple change '>='
for '>', she/he is the Author, she/he has improved their stats, nobody
knows/cares about the QA person work.

And the conclusion is that QA must go behind devs begging for I don't know
what, I have no words.

Maybe will be better if you and who thinks like you, rethink about how the
people can approach to the QA, without the option to feel that they are
there only to clean the basket.

>
> Best,
> Joel

But I don't think into give up, fortunately some things are changing, as a
much more significant attention to the help or the interesting discussions
about more and better regular test, or even easier to see taking a look to
last minutes of ESC.  I'm feeling a bit happier. 

I hope sooner rather than later we find the space to propose achieve an ISO
certification in a few years. What really would be a key for open a lot of
doors.

Miguel Ángel



--
View this message in context: 
http://nabble.documentfoundation.org/Libreoffice-qa-Reminder-QA-Meeting-on-Wednesday-tp4167253p4167534.html
Sent from the QA mailing list archive at Nabble.com.
___
List Name: Libreoffice-qa mailing list
Mail 

Re: [Libreoffice-qa] Reminder: QA Meeting on Wednesday

[Pedro]

Hi Joel, Cor, all


jmadero wrote
> I would like to take part also - I've thought about a few new ideas that
> I think might work :) I won't be available until December 8. Almost any
> time (absent really really late to really really early) would work
> between December 8 - 10. Shall we aim for a phone call? Google hangout?
> IRC Chat?
> 
> I'm happy Pedro brought this up as I think collectively we can come up
> with a solution.

I am available in that period. Anybody who would like to contribute with
ideas is very welcome!
Personally I prefer IRC chat, since it is always available, it's low
bandwidth and can be used at any hour without disturbing/waking anyone. We
can probably meet in #libreoffice-qa

Best regards,
Pedro



--
View this message in context: 
http://nabble.documentfoundation.org/Libreoffice-qa-Reminder-QA-Meeting-on-Wednesday-tp4167253p4167535.html
Sent from the QA mailing list archive at Nabble.com.
___
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/