Vivek <vivek@collab.net> writes:
> > On Windows, 'svn log' ends lines with CRLF instead of just LF. Does
> > your comparison above still work then (and if so, why? :-) ) ?
> Wait, don't shoot me! I never test under Windows before sending patches.
> :-)
I usually can't either; in fact I recently broke a couple of tests
that way, thank goodness D.J. Heap was around to analyze the breakage.
Just try to do the best you can for platforms you can't test on.
> > This is interesting, I don't think we've had any automated hook
> > testing yet, but this strategy seems to work. Does it pass for you on
> > both Unix and Win32 systems?
>
> See my confession above. :-)
Heh. Well, I recommend not trying fancy stuff like hook invocation in
the regression tests when you can't test that it worked.
> > > + e_out, err = svntest.main.run_svn (1, 'cat', '-r',
> > > + '{' + str(arg) + '}', mu_path)
> > > + if err.pop() != 'svn: Bogus date\n':
> > > + raise svntest.Failure
> >
> > What does "e_out" signify, how is it different from "out"?
> Just another variable. Note that variable out is used later. To prevent
> out from being overwritten I introduced a new variable.
Ah. Okay. A descriptive var name might be better (like "ignored_out"
for example).
I see the new patch is posted, I'll go review now.
-Karl
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Wed Jul 6 21:36:28 2005