Re: [jgit-dev] Strange failing tests under Linux which go through when

[Christian Halstrick <christian.halstrick@xxxxxxxxx>]

Hi

finally I found it. This platform-depending, performance-depending
thing costed me 2 days. After adding a lot of logging I found that
when working with GitIndex it is decided when to re-read a index from
disc is decided purely on modification time (see
GitIndex#rereadIfNecessary()). In fast test's where we face the racy
git problem we don't read GitIndex from disk although it would be
required.
I'll fix the tests so that we have a running build again. Since
GitIndex is deprecated I would like to limit the amount of work I'll
spent to fix GitIndex().

Ciao
  Chris