- User Since
- Mar 5 2015, 12:44 PM (163 w, 5 d)
Sun, Apr 22
Add bug number
Improve unittest by actually testing the body too.
It looks like the issue is that forgetDirs() doesn't find smb:// on the line of code that says DirectoryDataHash::iterator dit = directoryData.find(urlStr);.
So it goes into early return and doesn't do dirData.listersCurrentlyHolding.removeAll(lister);
which later leads to using a deleted lister that is still in listersCurrentlyHolding (for a slightly different URL).
Sat, Apr 21
use QVERIFY instead of Q_ASSERT, as it was before
Fri, Apr 20
I meant this stuff:
I find this solution horrible. QPointers screams "we designed this wrongly, the ownership rules are unclear, stuff gets deleted randomly, we have to guard against it".
Looks good, but shouldn't you remove the Q_PRIVATE_SLOT declarations too? They aren't used anymore, AFAICS.
Wed, Apr 18
kaddressbook does support cutting entities, that's in fact exactly what the initial bug report is about (which led to this code).
Tue, Apr 17
Mon, Apr 16
Signals having changed from protected to public in Qt5, the whole idea of preventing others to emit signals went down the drain anyway. Extending that to KJob we could make the signals non-private there, with strong documentation "don't emit this yourself".
Sat, Apr 14
Wordpress *and* ruby, way out of my comfort zone.
Can someone look into porting the current script?
git clone firstname.lastname@example.org:sysadmin/release-tools -b frameworks/5.0
then read create_sources_inc and look for svn
I think this patch makes sense, even though I'm sad that it doesn't actually fix the bug I was seeing (you call it broken config file, in that case that would mean that the code writing out that config file is -- or was -- broken, if it shouldn't write out ",Regular").
Wed, Apr 11
No. As soon as the RC1 tags are there, the message freeze is lifted.