Bug 564006

Summary:	dev-qt/qt-creator-3.5.1 crashes
Product:	Gentoo Linux	Reporter:	Fabien <fabieng>
Component:	Current packages	Assignee:	Davide Pesavento (RETIRED) <pesa>
Status:	RESOLVED UPSTREAM
Severity:	normal	CC:	qt
Priority:	Normal
Version:	unspecified
Hardware:	All
OS:	Linux
Whiteboard:
Package list:		Runtime testing required:	---
Attachments:	output of emerge --info backtrace

Description Fabien 2015-10-24 22:03:18 UTC

Created attachment 415390 [details]
output of emerge --info

Qt-Creator v3.5.1 (~amd64) is crashing when trying to lookup nested variables.

Since I updated to version 3.5.0, I noticed that Qt-creator is slower and uses more CPU. I need to wait 5-10s for the auto-completion while it was almost immediate before. It also segfaults from time to time, I haven't detected any pattern for the segmentations faults.

It should be noted that my program makes a quite intensive usage of templates through the eigen library (if anyone want to try, the code is here : https://bitbucket.org/specmicp/specmicp). 

I just want to be sure that the packaging hasn't change and that I'm not doing anything stupid before reporting this behaviour upstream.

Comment 1 Fabien 2015-10-24 22:05:38 UTC

Created attachment 415392 [details]
backtrace

This is the partial backtrace from gdb for one of the segfault.

It should be noted that the backtrace has more that 10000 frames, which may be one of the root of the problem...

Comment 2 Davide Pesavento (RETIRED) gentoo-dev

2015-10-26 18:25:44 UTC

(In reply to Fabien from comment #0)
> Since I updated to version 3.5.0, I noticed that Qt-creator is slower and
> uses more CPU. I need to wait 5-10s for the auto-completion while it was
> almost immediate before. It also segfaults from time to time, I haven't
> detected any pattern for the segmentations faults.

I also noticed a general slowdown in syntax highlighting and completion with 3.5.x, especially with template-heavy code (boost in my case). It's not as slow as you say, it takes at most 2-3 seconds here, but it's definitely noticeable compared to previous versions, and highly annoying.
OTOH, I have not observed an increased rate of crashes in 3.5.

> I just want to be sure that the packaging hasn't change and that I'm not
> doing anything stupid before reporting this behaviour upstream.

Nothing substantial has changed, unless you're using the clang code model, which is still experimental AFAIK and not recommended for "production" use.

So I don't think these (the slowness and the crashes) are downstream problems. I recommend to report them upstream. Thanks.