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

SOLVED: diff_tests 47 problem

From: Karl Fogel <kfogel_at_red-bean.com>
Date: Mon, 24 Mar 2008 12:43:12 -0400

Some of us have encountered an intermittent problem with diff_tests 47
('svn diff -x wc -r#' should return error). Today it happened to
Stefan Sperling, see http://paste.lisp.org/display/57924.

I think I know what's causing it, and am working on a fix. Just
posting here in case anyone else was about to start debugging. The
beginning of my log message explains all:

Fix an intermittent problem with diff_tests 47 ('svn diff -x wc -r#'
should return error), by moving some option-parsing code earlier.

The problem was that the RA session got opened before the diff options
were parsed. But since sbox.wc_dir occupied the position of a -x
extensions argument (that's the problem we're testing for), the test
would contact the repository from the regression suite's own working
copy -- that is, the Subversion repository itself. After contacting
that repository, the test would hit the expected option-parsing error
before it did any diffs, so as long as contacting svn.collab.net went
fine, you'd never know there was anything wrong. But if contacting
the repository failed (say, because of a cert error over https://),
then the test would fail, because it would get a different error than
it was expecting.

* subversion/libsvn_client/diff.c
  ### Make options get parsed before repository is contacted. ###
  ### (Slightly non-trivial, due to the way the code is arranged.) ###

To unsubscribe, e-mail: dev-unsubscribe_at_subversion.tigris.org
For additional commands, e-mail: dev-help_at_subversion.tigris.org
Received on 2008-03-24 17:43:21 CET

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