Libgit2: bump to a0ce87c51 #235
Conversation
|
@dahlbyk I cheated on my test. The answers were in #234. 📚 Hopefully there won't be too many merge conflicts; I only changed a couple of things. @nulltoken, let me know if this is satisfactory, and I'll squash it all into one commit. |
|
🆒 I will rebase #234 on this when it gets merged. |
| var name = Utf8Marshaler.FromNative(namePtr); | ||
| var value = Utf8Marshaler.FromNative(valuePtr); | ||
| return new ConfigurationEntry(name, value); | ||
| return Proxy.git_config_foreach(LocalHandle, (entry) => { |
There was a problem hiding this comment.
I don't believe this is correct - entry corresponds to a git_config_entry * on the native side, but GitConfigEntry is a struct on the managed side. So, what we are really getting here is a pointer to a git_config_entry, and not the git_config_entry structure itself.
I think you could either:
- Take an IntPtr here for entry and marshal the structure manually
- Declare GitConfigEntry as a class instead of a struct
I would prefer option 1.
|
I took the liberty of writing |
| public IntPtr valuePtr; | ||
| public uint level; | ||
| } | ||
| } No newline at end of file |
There was a problem hiding this comment.
Look Ma! A missing trailing new line :)
|
@ben Nice Job! TeamCity is smiling and so am I :) The two failing tests are the result of OpenSSL not being deployed yet (@dragan is aware of this). Could you please squash all the update-related commits so that I can update vNext and make @jamill happier? Regarding ddd41ba, I like it very much! Could you please attach it to #169 and turn the issue into a PR? I'd prefer to deal with this separately from the upgrade of libgit2 binaries. |
|
Squashed, and I trimmed out the script that updates libgit2 (that's over in #169 now). |
|
Re: |
So it'll have the progress-callback hotness!
Here's what happened in order to fix the tests:
GIT_STATUS_*enums. Their names are still bad, and they don't have comment docs.git_remote_download's signaturegit_config_entrystructure 🚧