On Thu, 2011-01-06, Prabhu Gnana Sundar wrote:
> I made a tabular summary of the change that this patch would make.
> Since it displayed weirdly in the mail body I have attached it as a
> file(diff-explanation-table.txt) with this mail.
> I have also attached the experiment carried out, in another
> file(diff-explanation.txt), with this mail.
> Please let me know if I am not clear at any point.
The tabular format is good but it would be easier to follow if instead
of "A" or "B" or "C" etc. you write "Shown as diff against source" or
"Shown as all lines added".
Why are some of your tests testing deletes, not adds?
$ svn diff -r2:1
--- copiedfile1 (revision 2)
+++ copiedfile1 (revision 1)
@@ -1,5 +0,0 @@
I seem to recall that the result sometimes depends on whether the target
of the diff command is the actual file or a directory that contains the
file. In other words, "svn diff ./" may behave differently from "svn
diff copiedfile1". Can you test that too please?
> > Are there any differences with different (old) versions of Subversion
> > server? (I seem to recall that copy-from data was not always supplied,
> > but I am not sure of the details.)
> Actually, the send_copyfrom_args was passed even without this patch, but
> was not handled. Only for 'svnserve' this was newly introduced by this
> patch. So I don't think there's something to worry.
> > Isn't it a layering violation for libsvn_wc to know about libsvn_ra?
> > Maybe this needs to use callbacks or something, so that all the RA
> > knowledge remains in libsvn_client.
> What you say is correct. :)
> But since we need to access a file which is not in our text-base. I had
> to get the ra_session to make use of the repository to fetch the file.
> Anyway now I am working on this to make use of any function in
> libsvn_client that would get me an ra_session. But that seems to me like
> a costly work.
> Thanks and regards
Received on 2011-01-06 15:23:21 CET