serf failures (was Re: [PROPOSAL][VOTE] Subversion)

[moving this back to dev]

Greg Stein wrote:
> On Tue, Nov 10, 2009 at 11:16, Blair Zajac <blair_at_orcaware.com> wrote:
>> On Nov 10, 2009, at 7:10 AM, Greg Stein wrote:
>>
>>> Regarding serf and Neon, I think that serf will be just fine to have
>>> as a default. It has been totally functional for many of us (cmpilato
>>> is a serf skeptic :-P)
>> Not yet though. It still fails in places that neon works.
>
> Anything besides 1.0 proxies?

Yes, two reproducible issues that do not involve a proxy. They both involve
"git svn" which use the Perl bindings. Maybe we want to keep that broken :)
This is with svn 1.6.5, git 1.6.5.2 and perl 5.8.9, all from a current MacPorts
build.

1) "git svn clone" fails with "Temp file with moniker 'svn_delta' already in use
  at /opt/local/lib/perl5/site_perl/5.8.9/Git.pm line 1022." This appears to be
  an error caused by a violation of the editor API.

Search for "editor" in this thread here:

http://n2.nabble.com/git-svn-fails-to-fetch-repository-td2151475.html

2) After getting a good "git svn clone" repository with Neon then fetching new
revisions with "git svn fetch" I get

$ git svn fetch
Malformed network data: XML parsing failed: (400 Bad Request) at
/opt/local/libexec/git-core/git-svn line 4618

I'm guessing it may be similar to this dump I emailed earier:

http://subversion.tigris.org/ds/viewMessage.do?dsMessageId=2411121&dsForumId=462

I can get a tcpdump for this case if people want one.

Blair

------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=462&dsMessageId=2418626
Received on 2009-11-16 19:20:14 CET