Contributing again

[Dag]

Hi, Friedrich.

Friedrich W. H. Kossebau skrev den 2016-07-01 13:20:
> Hi Dag,
> 
> happy to see you back and giving Plan (and hopefully more) some needed 
> love.
Thanks.
> 
> When no-one had been found to take over maintenance, I felt Plan was 
> some too
> good software to just wipe it, so took it as some playground and had 
> done some
> what I considered "cleaning" and also tried to do the port to Qt5/KF5 
> for it.
A lot of cleaning is very much needed, indeed. I will focus on the 
release for know but I have a few things I have thought about for some 
time we should maybe discuss a bit before you put too much effort into 
it.
> 
> So far I only roughly reached that porting goal. While everything 
> builds and
> works to some initial degree, there is at least one big issue:
> during the port from KLocale (which is deprecated and in 
> kdelibs4support) to
> Qt5's QDateTime with built-in timezone support I very possibly missed 
> some
> scenarios and have broken something. At least the unit tests when run 
> for
> timezones closer to the daybreak (e.g. Canada) begin to fail, where 
> they work
> fine in European timezones. Also do the schedulers not return the same 
> results
> as in Plan 2.9 IIRC. Had not yet found the time to tackle that.
Intriging behaviour, haven't looked at it yet, but as said in my other 
post maybe using time_t in place of KDateTime could be a solution.

> Then the support for currencies from KLocale is not (yet) covered by 
> Qt5, an
Shouldn't be a problem, see my other post.
> issue other projects like KMyMoney also face. No solution for that one 
> besides
> keeping use of KLocale. Just has the disadvantage that there is no 
> systemwide
> way to configure currency settings, as Plasma Systemsettings only 
> supports
> what is exposed in Qt. So default currency settings can be only taken 
> from the
> installed KLocale data. Not sure what to do here.
> Next to that there are some crashes due to libKGantt, but seems you 
> already
> investigated there and fixed first (or all?), great :)
Have not looked at KGantt yet, the fixes where for KChart.
> 
> Am Freitag, 1. Juli 2016, 11:10:03 CEST schrieb Dag:
>> Thanks, both.
>> Sure is glad you are still around, tackling words would be a bit
>> daunting ;)
>> 
>> I ran the tests in libs and got a couple of errors, e.g:
>> 
>> FAIL!  : TestXmlReader::testRootError() Compared values are not the 
>> same
>>     Actual   (errorMsg)                                    : "Ekstra
>> indhold sidst i dokumentet."
>>     Expected (QString("Extra content at end of document.")): "Extra
>> content at end of document."
>> 
>> I assume this is ok, just the returned errormsg is localized.
>> But, if I run in C locale, could this hide other problems?
>> I saw something in the git log about localized values in xml files.
> 
> Oh, interesting, I never hit that one, perhaps I am missing some Qt
> translation catalogs in my install (and KDE CI as well).
> This test should rather see fixing IMHO. Running in C locale might very 
> much
> hide other problems, actually at least in Sheets & Plan we currently 
> have some
> tests failing due to locale problems.
> 
>> and I get:
>> QFATAL : TestColorConversionSystem::testConnections() Test function
>> timed out
>> 
>> Is this real, or can it be that my system is missing something?
> 
> Timeout is real, yes. No idea yet what to do with too long running 
> tests in
> general (also an issue in Marble, Krita, etc.).
I tried to increase ctest --timeout to 500 seconds but that didn't make 
any difference so either its a different timeout, or its a bug in ctest.
Hmmm, maybe there is a way to send an "alive" signal to ctest?
> 
> I was to point to 
> https://build.kde.org/view/Calligra/job/calligra%20master
> %20kf5-qt5/PLATFORM=Linux,compiler=gcc/ for some reference when it 
> comes to
> failing tests (in Canadian/US timezone/locale) but these very minutes 
> that is
> done, hopefully back when you try the link.
> (last build was reported by CI as failed due to kreport build product 
> not
> being updated to latest kproperty, because CI does not track 
> inter-project
> dependencies on builds yet)
> 
> WRT review policy I agree with Camilla, review only needed for common 
> areas.
> For Plan, even while I have committed a lot there in the recent months, 
> I yet
> have no complete picture of all code, so I bow to you here and will be 
> happy
> to see your commits going straight in and learn from them, while myself 
> will
> now have any Plan commits of mine passed by your eyes first (none 
> planned the
> next days though).
> 
> I personally would be happy to see Plan being brought into a working 
> state
> again, so it can be part of the 3.0 release. By the questions on the 
> forums
> and bug reports I saw there are still some people interested in 
> something like
> it. Not everyone is sold to doing planning with Web interfaces :)
> 
> Cheers
> Friedrich
> _______________________________________________
> calligra-devel mailing list
> calligra-devel at kde.org
> https://mail.kde.org/mailman/listinfo/calligra-devel