- User Since
- Jan 25 2016, 10:25 AM (77 w, 6 d)
You can configure one in either the shortcut dialog or by right-clicking the menu entry, I use Meta+A ...
I agree with your thoughts in general, I just want to avoid breaking things and I'm really at loss about what this code is doing there ;)
seems fine now
Hmm, I'm still confused why this code was there then. Looking at git history, it was also changed several times, fixing issues. Does invoking the action via its shortcut still work?
Huh, does right-click -> git -> show annotation still work after that? If yes, there needs to be a copy of this code somewhere.
Fri, Jul 21
Thu, Jul 20
Wed, Jul 19
shoooooooould be fixed.
Tue, Jul 18
fixed now, but dragging below document end with triple-click doesn't select the last line.
Thu, Jul 13
Looks sensible to me but I'm not really familiar with the php support.
Wed, Jul 12
Oh, sorry then.
Tue, Jul 11
Hmm. Why not use dialog->setAttribute(Qt::WA_DeleteOnClose) instead? That seems to be the standard way to do it, no?
Thu, Jul 6
Yes, I felt responsible for fixing it after pushing the broken version and did that wrong, too :/ my bad. Thanks for the patch.
Tue, Jul 4
No, I was thinking that the next block with the openContext(...) should automatically re-use the internalContext() of the declaration if that is indeed the same. In any case, when looking at it like this, don't you need to set the new range on the context? Doesn't that become a problem when you extend the range during editing?
Looks sensible to me.
Mon, Jul 3
Can't spot anything obviously wrong -- go for it ;)
Sorry for taking so long!
Thu, Jun 29
Tue, Jun 27
Seems sensible to me from a first look, certainly more than it was before ;)
Mon, Jun 26
After some lookup I found that there are 2 declarations added somehow with same qualifiedIdentifier.
Also, looks like it my added test visitTypeSpec is somehow called twice thus in first call declaration is present once and it doesn't contain internal context and in second call there is 2 declarations where first one still doesn't have internal context and second have one.
I will have to look closer later, but typically the builders run twice per pass, and in the second pass the declaration created by the first pass should be *re-used*, i.e. openDeclaration() figures out which declaration was created by the first pass and re-uses that. It uses the scope identifier and the range to do that lookup. It might be that this fails. It definitely fails in some cases -- that is why your completion items sometimes disappear while the popup is visible.
Sun, Jun 25
From a first look it might be you can just import the first struct's context into the second one:
where first and second are the internalContext()s of the declarations declaring the two stucts. Maybe that already does what you need. Otherwise we can see :)
I wonder -- is this the right place for that logic? Does it work in highlighting, for example?
Jun 19 2017
The highlighting still looks wrong but that's a different issue. Patch looks ok -- it has a test, it can't be bad ;)
Jun 13 2017
Jun 3 2017
LGTM, I have actually seen crashes a few times which I would account to something related.
Sorry. I clicked around a bit with just one project open and it seems fine, I'll submit it to master, then we have enough time to catch issues. Thanks!
May 29 2017
Looks good to me, thanks. Should I push it or do you have commit access?
May 19 2017
Looks good, thanks! Useful change.
May 13 2017
May 5 2017
Yes, makes sense. Do you have commit access?
May 4 2017
May 3 2017
I think this is pretty much the sane solution to the problem ... still, we might want to turn it off for languages that don't need it, it seems like a non-negligible overhead if you consider a whole project. Maybe introduce a new normalizeLineEndings() member function which is called explicitly by the parse job ...?
May 1 2017
Apr 29 2017
Context? Reason? Effect?
Apr 28 2017
Apr 17 2017
Apr 14 2017
+1, sounds good to me
Looks fine to me! Thanks a lot once again :)
Apr 12 2017
I don't have a strong opinion on this, if you have, feel free to change it.
Mar 26 2017
Looks good except what is noted below, thanks!
Mar 25 2017
Looks good, please push :)
Mar 21 2017
Can you make the line edit disabled when something else than "custom command" is selected?
Otherwise, LGTM, thanks for the patch!
Mar 19 2017
Hm, I think the biggest problem is that we don't know how we want it to work ... so we can't write tests either ;)
Mar 16 2017
Sorry that my comment is a rather philosophical one, and maybe a bit disconnected from reality, but ...
Do we really need a full web browser engine for this purpose? Effectively, we render text, links, and maybe ocassionaly a simple table and an image. Isn't there something simpler than a web browser engine (which is half an operating system these days) which provides us with these capabilities? What about the Qt Rich Text stuff?