[MERGE] Misc fixes for win32

Andrew Bennetts andrew at canonical.com
Thu Jan 3 07:13:46 GMT 2008

Aaron Bentley wrote:
> Hi all,
> I decided to do some more work on win32.  The good news is that most of
> our test failures are problems in the tests, not problems in the code
> being tested.
> In particular, it appears that it is not safe to call
> WorkingTree.basis_tree() on an unlocked tree.  If this is done, later
> calls to WorkingTree may attempt to lock the tree, causing lock contention.
> Unfortunately, the HTTP tests are a real mess.  First of all, we're
> getting socket errors, and if I handle the socket errors, then it tries
> to do thread.join, and that blocks forever.  I don't have a lot of
> experience with threads, so I hope someone else can have a look at this.
> I had to make a few changes to the smart server protocol, because it
> makes some incorrect assumptions.  It assumes that the only outcome from
> doing get_files_bytes on a directory is ReadError, but on win32 (and
> possibly Windows-based servers), you get PermissionDenied.

I probably should have added PermissionDenied to the protocol sooner.  Thanks
for adding that in.

The other changes seem fine to me, and it sounds like Vincent has tracked down
the HTTP problems, so:



More information about the bazaar mailing list