Revision graph broken (regression)

[pmiossec]

Environment

• GitExtensions version: master ( 0b74e26 )

Issue description

At least on repository opening (I don't know if it could happen during a refresh)
it seems only when the repo has unrelated histories (but not sure)
the graph is broken:

instead of expected:

Refreshing the repository seems to make the bug disappear...

Steps to reproduce

No clear steps:

1. switching from repos with unrelated histories increase the probability to get the bug

Did this work in previous version of GitExtensions?

Bisected to 609e03d, so yes, before

Diagnostics

Handling of e.Handled is too aggressive?

[pmiossec]

@mstv I can make a PR to revert to my previous version of the optimization that don't introduce the regression.
Or you would like to investigate?

[mstv]

I have seen similar artifacts, too, mostly after rebase operations. I had two suspects: the commit identified by you or #11205.

I can make a PR to revert to my previous version of the optimization that don't introduce the regression. Or you would like to investigate?

That variant had other issues.

I think this will be resolved by the planned graph rendering in the background. Give me a few weeks for this rewrite, please.
If you think, the issue is too severe, just revert the commit.

[mstv]

Try it with #11718 and #11719, please.

[pmiossec]

Try it with #11718 and #11719, please.

I think it fixes the issue. Well done.