I just spent a day sorting this out and thought it worth mentioning
here. The scenario is as follows: I tried to do a merge between two
files in different parts of our repository into a working copy on my
machine. After reading the documentation about svn merge
(http://svnbook.red-bean.com/en/1.5/svn.ref.svn.c.merge.html as well
as "svn help merge"), I was under the (mistaken, as it turns out)
impression that WCPATH had to be the root of my working copy.
To make an example, in the repository I have:
The file in directoryA.2/bin/file is an svn copy of directoryA.1/bin/file.
On my machine, I've checked out directoryA.1 as "WC_directoryA.1" and
directoryA.2 as "WC_directoryA.2". I applied a change to "file" in
the working copy WC_directoryA.1, and committed. Now, I want to merge
that into my checkout in WC_directoryA.2.
If I do "svn merge --dry-run
inside WC_directoryA.2, it fails with this error message:
svn: REPORT request failed on '/svn/pipeline/!svn/vcc/default'
svn: REPORT of '/svn/pipeline/!svn/vcc/default': Could not read chunk
size: Secure connection truncated
This happened with versions 1.4.4 and 1.6.1 (although with 1.6.1 the
error message consisted of only the second of the two lines reported
The error message lead me to search the cause of this error in our
apache installation, our SSL setup, etc. Then, accidentally, I tried
the same command from inside WC_directoryA.2/bin, and this time it
So, what I am trying to say is, that the error was obviously mine for
misunderstanding what the documentation meant by "WCPATH", but the
error message I got from running svn lead me to search for the mistake
in an entirely wrong direction.
I think it would be useful to get a more telling error message in this
scenario, something along the lines "cannot find file to path in given
WCPATH" or so.
With kind regards,
Received on 2010-01-08 15:45:30 CET