Linus TorvaldsIn the current git repository, try this:
gitk origin/master..origin/pu
and it starts out looking ok.
But then scroll down a bit (say, press "PageDown" ten to twenty times), and scroll up again, and now that thing is totally unusable.
Is it just me?
Linus
Johannes SchindelinRe: gitk layout bug
Hi,
On Sat, 19 Jan 2008, Linus Torvalds wrote:
> In the current git repository, try this: > > gitk origin/master..origin/pu > > and it starts out looking ok. > > But then scroll down a bit (say, press "PageDown" ten to twenty times), > and scroll up again, and now that thing is totally unusable. > > Is it just me?
Just tried; I'm getting a lot of horizontal pink lines which seem to extend infinitely to the right, where at first they were vertical lines...
So it is not just you, I guess.
Ciao, Dscho
Alex RiesenRe: gitk layout bug
Linus Torvalds, Sun, Jan 20, 2008 02:44:58 +0100:
> > In the current git repository, try this: > > gitk origin/master..origin/pu > > and it starts out looking ok. > > But then scroll down a bit (say, press "PageDown" ten to twenty times), > and scroll up again, and now that thing is totally unusable. > > Is it just me?
No, it was already reported:
Message-ID: <20071104104618.GA3078@steel.home>
Paul MackerrasRe: gitk layout bug
Linus Torvalds writes:
> In the current git repository, try this: > > gitk origin/master..origin/pu > > and it starts out looking ok. > > But then scroll down a bit (say, press "PageDown" ten to twenty times), > and scroll up again, and now that thing is totally unusable. > > Is it just me?
No, I have seen it occasionally (though not recently), but I've never been able to trigger it when I had some time to debug it. :/
I'd be interested to know if you can trigger it with the "dev" branch of gitk, too. The dev branch doesn't use --topo-order or --date-order, but instead does the topological sort itself, so the cold-cache case is a lot faster (and the hot-cache case goes just as fast as before). And because I can now reorder the commits, I can put each open-circle commit just after the last merge that has it as a parent, which ends up looking much nicer too.
Paul.
Johannes SixtRe: gitk layout bug
Paul Mackerras schrieb:
> Linus Torvalds writes: > >> In the current git repository, try this: >> >> gitk origin/master..origin/pu >> >> and it starts out looking ok. >> >> But then scroll down a bit (say, press "PageDown" ten to twenty times), >> and scroll up again, and now that thing is totally unusable. >> >> Is it just me? > > No, I have seen it occasionally (though not recently), but I've never > been able to trigger it when I had some time to debug it. :/ > > I'd be interested to know if you can trigger it with the "dev" branch > of gitk, too. The dev branch doesn't use --topo-order or > --date-order, but instead does the topological sort itself, so the > cold-cache case is a lot faster (and the hot-cache case goes just as > fast as before). And because I can now reorder the commits, I can put > each open-circle commit just after the last merge that has it as a > parent, which ends up looking much nicer too.
The dev branch is fine in this regard. In fact, this is the primary reason why I'm using it.
BTW, there are still a few layout glitches, in particular, when F5 (Update) is used after frequent rebase/cherry-pick/commit/--amend. I'll bug you once I know how to reproduce them.
-- Hannes
Alex RiesenRe: gitk layout bug
Paul Mackerras, Mon, Jan 21, 2008 10:50:34 +0100:
> I'd be interested to know if you can trigger it with the "dev" branch > of gitk, too. The dev branch doesn't use --topo-order or > --date-order, but instead does the topological sort itself, so the > cold-cache case is a lot faster (and the hot-cache case goes just as > fast as before). And because I can now reorder the commits, I can put > each open-circle commit just after the last merge that has it as a > parent, which ends up looking much nicer too.
It is better, but sometimes the layout seem to be a bit too "aggressively cleaned up": there are just merges and arrows pointing in all directions but very few of real information.