threads / discuss / 10638

gitk graph routing problem

Subject: gitk graph routing problem

## tl;dr

8 messages between Nov 4, 2007 and Dec 14, 2007.

replies: 7people: 3as markdown or json

Alex Riesen· Nov 4, 2007, 10:46 UTC · lore

To reproduce, try running in git repo:

    gitk 02f630448e5d48e..06ea6ba9cf46ef5

Than go some pages (around 5) forward. You should notice system load going up rapidly. Now try paging back - and graph starts stretching to the right, to the point nothing fits on the screen anymore.

Gitk as of commit 7388bcbc5431552718dde5c3259d861d2fa75a12 in git repository. Git - v1.5.3.5-551-g02f6304 (terse fetch output and some *really* unrelated local modifications. Anyway, the gitk is the same). Bisect points to or around 6e8c87070306a757c4d7fd2c55cca3a90fe140c7 "gitk: Establish and use global left-to-right ordering for commits". It loops heavily at this commit and looks broken in all subsequent commits.

Last know good was 3244729aac7515c. The master of gitk repo is ok too.

Paul Mackerras· Nov 4, 2007, 21:47 UTC · re: Alex Riesen · lore

Re: gitk graph routing problem

Alex Riesen writes:

> To reproduce, try running in git repo:
> 
>     gitk 02f630448e5d48e..06ea6ba9cf46ef5

I can't reproduce it here, as my clone of the git repo doesn't have 02f630448e5d48e in it, even after updating...

Paul.

Junio C Hamano· Nov 4, 2007, 23:14 UTC · re: Paul Mackerras · lore

Re: gitk graph routing problem

Paul Mackerras <paulus@samba.org> writes:

Show 8 quoted lines

> Alex Riesen writes:
>
>> To reproduce, try running in git repo:
>> 
>>     gitk 02f630448e5d48e..06ea6ba9cf46ef5
>
> I can't reproduce it here, as my clone of the git repo doesn't have
> 02f630448e5d48e in it, even after updating...

Heh, me neither ;-).

Alex Riesen· Nov 5, 2007, 00:53 UTC · re: Junio C Hamano · lore

Re: gitk graph routing problem

Junio C Hamano, Mon, Nov 05, 2007 00:14:29 +0100:

Show 12 quoted lines

> Paul Mackerras <paulus@samba.org> writes:
> 
> > Alex Riesen writes:
> >
> >> To reproduce, try running in git repo:
> >> 
> >>     gitk 02f630448e5d48e..06ea6ba9cf46ef5
> >
> > I can't reproduce it here, as my clone of the git repo doesn't have
> > 02f630448e5d48e in it, even after updating...
> 
> Heh, me neither ;-).

Stupid me. Will follow up with a link to the archived repo in private mail.

Alex Riesen· Nov 5, 2007, 01:17 UTC · re: Alex Riesen · lore

Re: gitk graph routing problem

Alex Riesen, Mon, Nov 05, 2007 01:53:49 +0100:

Show 16 quoted lines

> Junio C Hamano, Mon, Nov 05, 2007 00:14:29 +0100:
> > Paul Mackerras <paulus@samba.org> writes:
> > 
> > > Alex Riesen writes:
> > >
> > >> To reproduce, try running in git repo:
> > >> 
> > >>     gitk 02f630448e5d48e..06ea6ba9cf46ef5
> > >
> > > I can't reproduce it here, as my clone of the git repo doesn't have
> > > 02f630448e5d48e in it, even after updating...
> > 
> > Heh, me neither ;-).
> 
> Stupid me. Will follow up with a link to the archived repo in private mail.
>

http://ftp-tmp:ftp-tmp@h877881.stratoserver.net/~ftp-tmp/break-gitk.git.tar.bz2

Its around 60mb.

Alex Riesen· Nov 5, 2007, 01:19 UTC · re: Alex Riesen · lore

Re: gitk graph routing problem

Alex Riesen, Mon, Nov 05, 2007 02:17:00 +0100:

Show 17 quoted lines

> Alex Riesen, Mon, Nov 05, 2007 01:53:49 +0100:
> > Junio C Hamano, Mon, Nov 05, 2007 00:14:29 +0100:
> > > Paul Mackerras <paulus@samba.org> writes:
> > > 
> > > > Alex Riesen writes:
> > > >
> > > >> To reproduce, try running in git repo:
> > > >> 
> > > >>     gitk 02f630448e5d48e..06ea6ba9cf46ef5
> > > >
> > > > I can't reproduce it here, as my clone of the git repo doesn't have
> > > > 02f630448e5d48e in it, even after updating...
> > > 
> > > Heh, me neither ;-).
> > 
> > Stupid me. Will follow up with a link to the archived repo in private mail.
> >

and of course, I send it to everyone... Ach, well. Will keep it for a week, or until it is reproduced somewhere else. Whichever comes first.

Paul Mackerras· Dec 11, 2007, 22:26 UTC · re: Alex Riesen · lore

Re: gitk graph routing problem

Alex Riesen writes:

Show 7 quoted lines

> To reproduce, try running in git repo:
> 
>     gitk 02f630448e5d48e..06ea6ba9cf46ef5
> 
> Than go some pages (around 5) forward. You should notice system load
> going up rapidly. Now try paging back - and graph starts stretching
> to the right, to the point nothing fits on the screen anymore.

I finally got back to look at this. The problem is not so much the layout algorithm per se as the fact that I haven't worked out a good way to pack lots of downward-pointing arrows in without using up arbitrarily large amounts of horizontal space. You have managed to find an example where just about every commit is a merge needing one or more downward-pointing arrows.

Incidentally, gitk from the dev branch of my gitk.git repo does much better on this example, since it is able to hoist the open-circle (excluded) commits up to the row below their merge children, which looks much nicer.

Paul.

Alex Riesen· Dec 14, 2007, 22:29 UTC · re: Paul Mackerras · lore

Re: gitk graph routing problem

Paul Mackerras, Tue, Dec 11, 2007 23:26:10 +0100:

Show 21 quoted lines

> Alex Riesen writes:
> 
> > To reproduce, try running in git repo:
> > 
> >     gitk 02f630448e5d48e..06ea6ba9cf46ef5
> > 
> > Than go some pages (around 5) forward. You should notice system load
> > going up rapidly. Now try paging back - and graph starts stretching
> > to the right, to the point nothing fits on the screen anymore.
> 
> I finally got back to look at this.  The problem is not so much the
> layout algorithm per se as the fact that I haven't worked out a good
> way to pack lots of downward-pointing arrows in without using up
> arbitrarily large amounts of horizontal space.  You have managed to
> find an example where just about every commit is a merge needing one
> or more downward-pointing arrows.
> 
> Incidentally, gitk from the dev branch of my gitk.git repo does much
> better on this example, since it is able to hoist the open-circle
> (excluded) commits up to the row below their merge children, which
> looks much nicer.

could you point to a specific commit where it does that?

Because the current head of dev branch at //git.kernel.org/pub/scm/gitk/gitk (3de07118f0993e6f7bc7ce02276751795d80b877) does not look any different and still almost locks up drawing all the horizontal lines.