https://bugs.kde.org/show_bug.cgi?id=379219
Lassi Väätämöinen changed:
What|Removed |Added
CC||rjvber...@gmail.com
--- Comment #34 from
https://bugs.kde.org/show_bug.cgi?id=379219
Milian Wolff changed:
What|Removed |Added
Status|REOPENED|RESOLVED
Resolution|---
https://bugs.kde.org/show_bug.cgi?id=379219
--- Comment #32 from Lassi Väätämöinen ---
(In reply to Sven Brauch from comment #30)
> You can turn off the git plugin in the options, no?
This is a viable workaround, but still inconvenient, as for example using the
'Annotation'/git blame
https://bugs.kde.org/show_bug.cgi?id=379219
Lassi Väätämöinen changed:
What|Removed |Added
CC||lassi.vaatamoi...@gmail.com
https://bugs.kde.org/show_bug.cgi?id=379219
Eugene Shalygin changed:
What|Removed |Added
CC||eugene.shalygin+bugzilla.kd
https://bugs.kde.org/show_bug.cgi?id=379219
--- Comment #31 from mrnuke ---
I had totally missed that. Thank you for the suggestion!
--
You are receiving this mail because:
You are watching all bug changes.
https://bugs.kde.org/show_bug.cgi?id=379219
--- Comment #30 from Sven Brauch ---
You can turn off the git plugin in the options, no?
--
You are receiving this mail because:
You are watching all bug changes.
https://bugs.kde.org/show_bug.cgi?id=379219
--- Comment #29 from mrnuke ---
Can I have my checkbox to disable git integration completely? Please?
I'll do my own git management rather than have sensitive operations, like
rebase fail. Thank you!
--
You are receiving this
https://bugs.kde.org/show_bug.cgi?id=379219
--- Comment #28 from Sven Brauch ---
Good to know somebody is considering this problem upstream ...
Yes, possibly we should provide a workaround on our side until then :/
So long, what you can do is pkill -STOP kdevelop and then
https://bugs.kde.org/show_bug.cgi?id=379219
--- Comment #27 from Daniel Santos ---
There is some hope coming around the corner:
https://public-inbox.org/git/20170921043214.pyhdsrpy4omy5...@sigill.intra.peff.net
It's not yet committed, but this bug may end up either
https://bugs.kde.org/show_bug.cgi?id=379219
--- Comment #26 from Daniel Santos ---
(In reply to Sven Brauch from comment #23)
I understand your position that git may want to look at this. Has there been a
bug report filed with them yet? I also think that I understand
https://bugs.kde.org/show_bug.cgi?id=379219
--- Comment #25 from Sven Brauch ---
I don't think that is a good comparison.
--
You are receiving this mail because:
You are watching all bug changes.
https://bugs.kde.org/show_bug.cgi?id=379219
--- Comment #24 from mrnuke ---
I think the comment that this issue isn't specific to KDevelop is misleading.
Yes, it's possible to type 'git status' while a rebase is in progress, the same
way it's possible to type 'rm -rf ~'. If
https://bugs.kde.org/show_bug.cgi?id=379219
--- Comment #23 from Sven Brauch ---
I don't even think this issue is specific to KDevelop. For example I have a
shell which does "git status" and shows a prompt depending on that -- lots of
people have that. If you press Enter
https://bugs.kde.org/show_bug.cgi?id=379219
--- Comment #22 from Daniel Santos ---
Sorry about my duplicate bug.
This really is a serious problem, you just *can't* interfere with source
control. I would even suggest working with the git team to see if git can add
(or
https://bugs.kde.org/show_bug.cgi?id=379219
Sven Brauch changed:
What|Removed |Added
Status|RESOLVED|REOPENED
https://bugs.kde.org/show_bug.cgi?id=379219
Sven Brauch changed:
What|Removed |Added
CC||daniel.san...@pobox.com
---
https://bugs.kde.org/show_bug.cgi?id=379219
--- Comment #19 from Wei-Cheng Pan ---
For what it's worth, `git status --porcelain` usually takes 0.3s for a clean
work copy on my machine.
--
You are receiving this mail because:
You are watching all bug changes.
https://bugs.kde.org/show_bug.cgi?id=379219
--- Comment #18 from Milian Wolff ---
@mrnuke: how long does `git status --porcelain -- path/to/linux` take on your
machine? On mine it finishes within ~200ms, according to perf stat:
Performance counter stats for 'git status
https://bugs.kde.org/show_bug.cgi?id=379219
--- Comment #17 from Sven Brauch ---
Well programming is complicated sometimes ...
"don't run any git operations unless the window has focus" certainly isn't an
acceptable solution either way.
--
You are receiving this mail
https://bugs.kde.org/show_bug.cgi?id=379219
--- Comment #16 from mrnuke ---
In that case, you'd have to determine if the process holding the lock is still
active in order to avoid an infinite loop. Or timeout after a while, increasing
complexity for a very niche use case.
https://bugs.kde.org/show_bug.cgi?id=379219
--- Comment #15 from Sven Brauch ---
It wasn't, but that makes sense. It could still wait for a bit for the lock to
become available though.
--
You are receiving this mail because:
You are watching all bug changes.
https://bugs.kde.org/show_bug.cgi?id=379219
--- Comment #14 from mrnuke ---
Git needs to release the lock in order for operations such as conflict
resolution to not deadlock, or rebase with edit. Or a number of other things.
I'm sorry for assuming this was obvious.
--
You
https://bugs.kde.org/show_bug.cgi?id=379219
--- Comment #13 from Sven Brauch ---
You created half a page of analogies but none of them has an actual argument
which explains why it makes sense for git to release the lock during rebasing
and then fail when something else
https://bugs.kde.org/show_bug.cgi?id=379219
--- Comment #12 from mrnuke ---
NAK. That's not how git is designed to work. I can take a Ferrarri and drive it
on a dirt road. It coming back all dinged and scratched is not something
Ferrari has to fix.
The interaction issue
https://bugs.kde.org/show_bug.cgi?id=379219
--- Comment #11 from Sven Brauch ---
I still think this needs to be fixed in git, if you run a rebase and then "git
status" in a different terminal, you will have the same issue ...
--
You are receiving this mail because:
You are
https://bugs.kde.org/show_bug.cgi?id=379219
--- Comment #10 from Wei-Cheng Pan ---
BTW, this is what I get in the log:
kdevplatform.vcs: Execute dvcs command: "git status --porcelain --
/***/firefox/widget/windows/JumpListBuilder.cpp"
kdevplatform.vcs: Execute dvcs
https://bugs.kde.org/show_bug.cgi?id=379219
--- Comment #9 from Wei-Cheng Pan ---
I think it's because `git status --porcelain`, it will do something like
`open("/repo/.git/index.lock", O_RDWR|O_CREAT|O_EXCL|O_CLOEXEC, 0666)`,
according to strace.
--
You are receiving
https://bugs.kde.org/show_bug.cgi?id=379219
--- Comment #8 from mrnuke ---
Created attachment 105593
--> https://bugs.kde.org/attachment.cgi?id=105593=edit
strace, attached to already running kdevelop, then the rebase boogieman strikes
I think it's much easier for
https://bugs.kde.org/show_bug.cgi?id=379219
--- Comment #7 from Milian Wolff ---
you could try to strace kdevelop to see the git commands it executes and files
it creates/locks, something like the following could do the trick:
strace -e open,execve,flock -f -s 1000 kdevelop -s
https://bugs.kde.org/show_bug.cgi?id=379219
--- Comment #6 from mrnuke ---
Created attachment 105500
--> https://bugs.kde.org/attachment.cgi?id=105500=edit
stdio output of Kdevelop while problem manifests
I finally remembered to start Kdevelop from the terminal, so I got
https://bugs.kde.org/show_bug.cgi?id=379219
--- Comment #5 from Kevin Funk ---
+1, btw,
I'm not looking forward to another potentially crashy library incorporated into
KDevelop either. We should find a different solution (but first we need more
information).
--
You are
https://bugs.kde.org/show_bug.cgi?id=379219
--- Comment #4 from Sven Brauch ---
Also I'm not super excited about libgit2, it's another incredibly complex piece
of software running in-process and potentially crashing (ask the kate people).
Out-of-process looks messy but tends
https://bugs.kde.org/show_bug.cgi?id=379219
Sven Brauch changed:
What|Removed |Added
CC||m...@svenbrauch.de
---
https://bugs.kde.org/show_bug.cgi?id=379219
Kevin Funk changed:
What|Removed |Added
Status|UNCONFIRMED |RESOLVED
Resolution|---
https://bugs.kde.org/show_bug.cgi?id=379219
Wei-Cheng Pan changed:
What|Removed |Added
CC||legnale...@gmail.com
---
36 matches
Mail list logo