(This isn't strictly speaking a git2go bug, please feel free to close and move the discussion elsewhere)
I was investigating crashes in our binary which uses git2go, which I now think might have to do with #218. In that investigation, I produced a small Go program - ~50 lines of CGo calling libgit2 - that consistently panics due to heap corruption. It seems like CGo intends to support the calling sequence demonstrated, so I think the bug may be specifically with CGo and libgit2.
I've raised this on the golang-nuts mailing list and thought that it might be of interest here. The tarball with the code and a script to reproduce the failure is available for download hosted by the mailing list.
(This is on Linux x86_64, recent kernel, go 1.4.2, libgit2 v0.22.3)
(This isn't strictly speaking a git2go bug, please feel free to close and move the discussion elsewhere)
I was investigating crashes in our binary which uses git2go, which I now think might have to do with #218. In that investigation, I produced a small Go program - ~50 lines of CGo calling libgit2 - that consistently panics due to heap corruption. It seems like CGo intends to support the calling sequence demonstrated, so I think the bug may be specifically with CGo and libgit2.
I've raised this on the golang-nuts mailing list and thought that it might be of interest here. The tarball with the code and a script to reproduce the failure is available for download hosted by the mailing list.
(This is on Linux x86_64, recent kernel, go 1.4.2, libgit2 v0.22.3)