- User Since
- Mar 5 2015, 12:44 PM (215 w, 2 d)
I fixed them ;-)
Glad that the unittest caught an actual bug for once :-)
I'm talking about jobtest and kdirmodeltest regressing exactly in build 87, which is where this commit landed.
If you click on history for a given test you can see that those aren't sporadic failures: https://build.kde.org/view/OS%20-%20Windows/job/Frameworks/job/kio/job/kf5-qt5%20SUSEQt5.10/87/testReport/junit/projectroot/autotests/kiowidgets_kdirmodeltest/history/
Thu, Apr 18
No, it's pretty much the same. The garbage looks a bit different, but switching to another email and back fixes it, so no big deal. I thought this was what this whole effort was about, though :-)
Thanks, this fixes the plasma crash on resume.
Ah, great, thanks.
Wed, Apr 17
Patch looks good now ;-)
I like the idea.
Tue, Apr 16
This change also introduced regressions in two unittests: jobtest and kdirmodeltest. I'll let you look into those :-)
Sorry, I was faster.... :/
Nah, no C++17 here, the compiler is just confused because it doesn't know the types uid_t and gid_t.
Mon, Apr 15
Nice pun :-)
Sun, Apr 14
fix commit log
Can't remember what use case I had in mind back then... Probably someone asked me to implement this....
Urgh. Indeed. And looking around I find many inline virtuals in apparently public headers... http://www.davidfaure.fr/2019/inline_virtual_dtors.diff (though maybe some of these don't have d pointers at all...)
The double star looks very C-ish to me, but OK, I just realized that the recursive call itself would probably look more cumbersome, so I won't insist.
Oops indeed, I stopped working on the unittest when I managed to make it reproduce the crash. Thanks for integrating it and fixing it.
Yes, the pitfalls are well known.
- don't modify the container being iterated upon (from inside the loop)
- don't use qAsConst on temporaries
Improve regexp to avoid catching kbuildsycoca.cpp
I'd still prefer if such changes would come with a unittest, because it will be hard otherwise to immediately detect (while developing) that changes don't introduce any regressions.
I managed to write a unittest for this one, please integrate it into your commit:
Sat, Apr 13
I view the KF5 changelog as the list of things that can be useful to the users of the frameworks (i.e. application developers).
When we add API, fix a bug, or change dependencies, that's useful for them to know.
When we repair a unittest, fix typos in comments, port away from deprecated methods and so on, I don't see how that is useful for the application developers to know. To me it would just be noise in the changelog, it doesn't affect them.
Sure, in the long run it means we're maintaining the stuff and making sure it will still work with future versions of Qt, but they'll get notified of that when the time comes anyway.
GIT_SILENT is "trivial, don't look at the commit" -- at least it used to be. Not really applicable to commits such as this one....
I actually meant Alt+Home/End, because I thought this was "go to beginning/end of view" (itempotent, if you do it twice, the second time it does nothing).
Alt+Home, Alt+End ?
BTW I didn't go for "making intervalChecker() return a base class pointer" because of the code that says AkonadiServer::instance()->intervalChecker()->requestCollectionSync(collection);
Hi Dan, can you push this ASAP? My own work keeps touching lines that have referenced() in them, a good recipe for merge conflicts for either of us at some point.
It's common to say "I have 3 columns everywhere", that's what tree models look like.
And if you have zero rows somewhere, it doesn't matter how many columns you have, that's still "nothing".
FYI NO_CHANGELOG doesn't have to be in the first line (which would make noise in phab review titles for new requests, etc). It can be a line of its own, for example towards the end of the commit log.
Yep, but extracting changelogs from commit logs can be mostly automated, including filtering out many commits with the same first line.
It's what I do in firstname.lastname@example.org:sysadmin/release-tools branch frameworks/5.0 file parse_changelogs.pl.
cgiboudeaux: the Qt documentation says that foreach is deprecated, what do you suggest Laurent should add to the commit log? Just "it's deprecated", a copy/paste of the Qt docu, or https://www.kdab.com/goodbye-q_foreach/ ?
Thanks for this contribution.
port interval method to std::chrono
Port to chrono_literals, even if it makes QtCreator 4.8 barf on the syntax ;)
Fri, Apr 12
Thu, Apr 11
Port to std::chrono.