I have an SVN repository at approx rev 100k. It contains a web
applications, and also configuration files pertaining to different
clients. A recent decision has mandated that client configurations be
abstracted out of the existed repository and placed in a new
repository, all the while preserving the revision information. I've
have a lot of luck thus far, but have hit a wall that I'm not sure the
best way to deal with. Thus far, I've created a dump filter and run
it through svndumpfilter to include relevant paths.
Occasionally, a developer would merge a revision from the a trunk into
a branch. These cause revisions of type 'Node-action: change' , but
unfortunately the file that is to be changed does not exist. I am
fairly new to subversion, and I would think that the a corresponding
'Node-action: add' would be issued first.
For example, the following revision is the result of an svn merge
(with some sensitive text omitted):
brought (merged carefully from trunk) Vik's changes:
1. svn merge -r 10416:10417
2. svn merge -r 10438:10444
The error (caused by `svnadmin load repo < webapp-20080402.filtered.10450` :
<<< Started new transaction, based on original revision 10450
svnadmin: File not found: transaction '10449-1', path
* editing path :
As you can see, these revision is trying to change a file that does
not exist. Should there not be a previous add statement, with a
Node-copyfrom-path that copies the file to the branch first? This
file does not exist I don't understand how subversion would issue a
change to a file that was not yet brought into the branch. I also
checked the raw, unfiltered dumpfile for revisions of this type that
may have been omitted in the filtered dump file.
Any insight is greatly appreciated!
To unsubscribe, e-mail: users-unsubscribe_at_subversion.tigris.org
For additional commands, e-mail: users-help_at_subversion.tigris.org
Received on 2008-04-03 23:30:30 CEST