On MacOS, 1.7.2, git svn clone seems to fail silently

git svn clone -s https://gate.svn.sourceforge.net/svnroot/gate

This worked with a previous version of git, I am sorry to report that I don't know which one.

When I run it with GIT_TRACE=2, the trace ends with:

tace: built-in: git 'config' 'svn-remote.svn.branches-maxRev' '12912' trace: built-in: git 'config' 'svn-remote.svn.tags-maxRev' '12912' trace: built-in: git 'config' 'svn-remote.svn.branches-maxRev' '12912' trace: built-in: git 'config' 'svn-remote.svn.tags-maxRev' '12912' trace: built-in: git 'gc' '--auto'

I get none of the usual 'checked through' output, either.

that's all. No errors, no messages. No branches, no contents. Just the wind whistling through the bytes.

Hi Benson,

Benson Margulies wrote:

[...]

Heh - you might be the one to figure out http://thread.gmane.org/gmane.comp.version-control.git/134936

But first, it would be good to confirm:

 - if you run "echo $?" right after such a failure, what
   status code do you get?

 - what is the output of "git branch -r" when you finish?

Coping strategy: usually with "git svn clone" failures, you can resume where you left off with "git svn fetch".

Good luck, Jonathan

On Fri, Jul 30, 2010 at 9:13 PM, Jonathan Nieder <jrnieder@gmail.com> wrote:

 I don't have $? at hand, but I can tell you that git branch -r says
 nothing, either. Also, running git fetch has the same result. Prints
 the same sort of trace lines if tracing, then does nothing

Hi again,

Benson Margulies wrote:

Yes, this seems different from the EPIPE bug, alas.

I get a core dump (for some small process, I don’t know what: 1048576 bytes. Died of signal 5 in pthread_mutex_unlock) and git-svn is still going.

 $ git svn clone -s https://gate.svn.sourceforge.net/svnroot/gate
 Initialized empty Git repository in /tmp/gate/.git/

Sitting there quietly.  ltrace tells me it’s busy copying around data.

 memcpy(0x0aa679d8, "Git::SVN::Ra", 12)                                 = 0x0aa679d8
 pthread_getspecific(0, 0xbfb1855c, 1, 0xb76b83a0, 0xaa679d8)           = 0x9df2008
 realloc(0x0aa679d8, 20)                                                = 0x0aa679d8
 memmove(0xaa679e4, 0x8160163, 7, 0xab723e8, 0xa5ae200)                 = 0xaa679e4
 memcpy(0xbfb186dc, "Git::SVN::Ra::SUPER", 19)                          = 0xbfb186dc
 memcpy(0xbfb185fc, "Git", 3)                                           = 0xbfb185fc
 memcpy(0xbfb185fc, "SVN", 3)                                           = 0xbfb185fc
 memcpy(0xbfb185fc, "Ra", 2)                                            = 0xbfb185fc
 memcpy(0xbfb185fc, "SUPER", 5)                                         = 0xbfb185fc
 memmove(0xa89f878, 0xa88f934, 19, 0x80d171d, 0x9df2008)                = 0xa89f878
 memmove(0xa89f88b, 0x8160032, 2, 0x80ecf3a, 0xa89f878)                 = 0xa89f88b
 memmove(0xa89f88d, 0xa257f7f, 10, 0x80ecf3a, 0xa89f878)                = 0xa89f88d
 memcpy(0x0a10cbe0, "\3701%\n\0302%\n82%\nX3%\n", 16)                   = 0x0a10cbe0
 strlen("Git::SVN::Ra")                                                 = 12
 memmove(0xa89f780, 0x9e16d7c, 12, 0x80d5cd1, 0xa10cbe0)                = 0xa89f780
 memmove(0xa89f798, 0xa89f780, 12, 0x8083aa6, 0x9df2008)                = 0xa89f798
 memmove(0xa89f8a8, 0xa89f878, 31, 0x8083aa6, 0x9df2008)                = 0xa89f8a8
 memcmp(0xa59c590, 0xa89f8bc, 2, 0x45505553, 0x93a3a52)                 = -1
 memcmp(0xa59c590, 0xa89f8bb, 2, 0x45505553, 0x93a3a52)                 = 0
 memcpy(0x0a4ee938, "\0302%\n82%\nX3%\n", 12)                           = 0x0a4ee938
 strcmp("REF", "SVN::Pool")                                             = -1
 strlen("SVN::Pool")                                                    = 9
 strcmp("SCALAR", "_p_apr_pool_t")                                      = -1
 strlen("_p_apr_pool_t")                                                = 13
[etc]

(strace tells me it’s in conversation with the server.) Hints?

I didn't realize that this email was the same as the gmane group and I thus made an unhelpful reply to the other thread you pointed me at. I apologize for the noise.

$? -> 0

The numbers in tags-maxRev don't go up with each 'fetch'.

On Fri, Jul 30, 2010 at 22:28, Benson Margulies <bimargulies@gmail.com> wrote:

Are you fetching to an already-existing repo? If you are, and it fails relatively quickly, it'd be helpful if you could upload the repo somewhere, so people can quickly test things (when I reproduced the bug on Linux on a new clone, it took tens of minutes before failing).

Hello everyone.

There's a problem here, but not even remotely (pardon) the one I thought it was.

The right URL in this case with -s is https://gate.svn.sourceforge.net/svnroot/gate/gate.

Note the extra /gate on the end.

It would be nice if this was diagnosed in fashion ("Dear user. That doesn't look like a stdlayout svn directory) but the injury is mos def self-inflicted.