[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Problem commiting - failed commit actually worked

From: Karl Auer <kauer_at_biplane.com.au>
Date: 2006-01-23 02:01:19 CET

I am experimenting with vendor branches and followed the plan in the SVN book. The initial import worked fine. I copied the current version to a tag version, and checked out the current version. I copied the next version over the current version in my working directory, and went to do a commit. The commit failed as in the next three lines:

    Transmitting file data ...
    svn: Commit failed (details follow):
    svn: Cannot connecto to host 'svn.id.ethz.ch'

Hm. OK, probably a network failure, my link is very slow. Oh, there's a typo in the error message too, but that's not so important. Anyway, I did another commit. This time it failed like this (line breaks added for clarity):

commit -m "Upgrade to dnsjava 2.0.0." /home/kauer/eclipse_workspace/current/doc/org/xbill/DNS/Generator.html
          [... lots more files on this one line...]
          /home/kauer/eclipse_workspace/current/USAGE
    Sending /home/kauer/eclipse_workspace/current/Changelog
    svn: Commit failed (details follow):
    svn: CHECKOUT request failed on '/jumbo/!svn/ver/241/kom/dk/vendors/dnsjava/current/Changelog'
svn: The version resource does not correspond to the resource within the transaction.
     Either the requested version resource is out of date (needs to be updated), or
     the requested version resource is newer than the transaction root (restart the commit).

I've read this error message a dozen times and I still can't get my head around the difference between a "version resource" and the "resource within the transaction".

However, looking at the repository, it seems that first commit actually worked! The version I have in my working directory is out of date, because *it* thinks the commit failed. So I would appear to have received a "commit failed" message when a commit did in fact succeed. I guess that's better than getting "commit succeeded" when the commit actually failed, but it's still not good.

It seems to me that failures of that type ("Cannot connect" or whatever) should return a much mre ambivalent message; the client had sent all it needed to send, but never received an acknowlegement. The error should have been something like "all data sent, but cannot confirm that commit succeeded".

Regards, K.
 
PS: This is using subclipse/JavaSVN. I guess it could all be a JavaSVN problem rather than a subversion problem.

-- 
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Karl Auer (kauer@biplane.com.au)                   +61-2-64957160 (w/h)
http://www.biplane.com.au/~kauer/                  +61-428-957160 (mob)
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Mon Jan 23 02:02:13 2006

This is an archived mail posted to the Subversion Users mailing list.

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.