From: Johannes Sixt <j6t@kdbg.org>
To: Michael Warmuth-Uhl <michael.warmuth-uhl@suljee.de>
Cc: git@vger.kernel.org
Subject: Re: gitk: race condition when changing view
Date: Sat, 18 Apr 2026 19:48:59 +0200 [thread overview]
Message-ID: <ee2599a5-13b6-469a-a284-4a5d26b2c721@kdbg.org> (raw)
In-Reply-To: <20260416233639.32e5b481@rosa.suljee.de>
Am 16.04.26 um 23:36 schrieb Michael Warmuth-Uhl:
> when displaying a bigger commit and then changing the view, gitk reports
>
> can't read "treediffs(b107a60c5bed3002ecef93b4405a44d809776e6f)": no such variable
> can't read "treediffs(b107a60c5bed3002ecef93b4405a44d809776e6f)": no such variable
> while executing
> "lsearch -exact $treediffs($ids) $fname"
> (procedure "makediffhdr" line 7)
> invoked from within
> "makediffhdr $fname $ids"
> (procedure "parseblobdiffline" line 56)
> invoked from within
> "parseblobdiffline $ids $line"
> (procedure "getblobdiffline" line 16)
> invoked from within
> "getblobdiffline file6 b107a60c5bed3002ecef93b4405a44d809776e6f"
> ("eval" body line 1)
> invoked from within
> "eval $script"
> (procedure "dorunq" line 11)
> invoked from within
> "dorunq"
> ("after" script)
>
> It is reproducible with:
>
> TCDIR=/tmp/testcase
> N=512
>
> rm -rf $TCDIR
> git init $TCDIR
> for i in `seq $N` ; do
> dd if=/dev/random status=none bs=1024 count=16 | uuencode "" > $TCDIR/foo_$i
> done
> git -C $TCDIR add .
> git -C $TCDIR commit -m "commit"
>
> cd $TCDIR
> gitk --all
>
> followed by an immediate change to "All Files" in gitk menu "View".
>
> That happens because proc showview (and some others) clear treediffs which is
> still needed by tasks pending in runq.
>
> The patch below makes showview wait until runq is empty which is slow (waits for
> useless operations) and feels wrong. I have no idea how to cleanly flush runq
> instead.
>
> The same issue has been mentioned here before:
> https://stackoverflow.com/questions/17977996/gitk-cant-read-treediffs-error
>
> Michael
Thank you for the report, the reproducer, and for digging into the issue.
Waiting for the process to complete is certainly not the most desirable
way to handle the situation. I've attempted to clear the run queue with
the patch below, but the reproduction recipe still triggers an error
'can't unset "currunq": no such variable', the reason being that proc
dorunq is entered recursively via one of the queued scripts, and I
haven't found why that is the case.
I'll look into this further as time permits, but don't hold your breath.
-- Hannes
diff --git a/gitk b/gitk
index dc042a5..51d9fe5 100755
--- a/gitk
+++ b/gitk
@@ -4693,6 +4693,17 @@ proc showview {n} {
}
unselectline
normalline
+ global runq
+ foreach curq $runq {
+ set fd [lindex $curq 0]
+ if {$fd ne {}} {
+ catch {close $fd}
+ } else {
+ set script [lindex $curq 1]
+ unset isonrunq($script)
+ }
+ }
+ set runq {}
unset -nocomplain treediffs
clear_display
if {[info exists hlview] && $hlview == $n} {
prev parent reply other threads:[~2026-04-18 18:36 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-04-16 21:36 gitk: race condition when changing view Michael Warmuth-Uhl
2026-04-18 17:48 ` Johannes Sixt [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=ee2599a5-13b6-469a-a284-4a5d26b2c721@kdbg.org \
--to=j6t@kdbg.org \
--cc=git@vger.kernel.org \
--cc=michael.warmuth-uhl@suljee.de \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox