Thanks for your feedback so far! I am currently working on the second part of the patch and running the test suite- took me a little time to get sorted out how to run it.
As for checking the log output, I could collect all the external start/end messages in a set and compare this against the ground truth set, without checking the order of them. What do you think? Cheers, Michael -----Ursprüngliche Nachricht----- Von: Philip Martin [mailto:philip.mar...@wandisco.com] Gesendet: Donnerstag, 13. November 2014 17:57 An: Julian Foad Cc: Schmidt, Michael; dev@subversion.apache.org Betreff: Re: [PATCH] issue #4527: notify start of exporting external Julian Foad <julianf...@btopenworld.com> writes: > Other than the indentation, Michael's patch looks like the right fix > in the right place, wouldn't you agree? Yes, I think it is correct. Extending the patch to fix the in_external nesting problem would be ideal but we could commit this version. > Michael, have you tested it by running the full test suite? If there > is a test that covers this then I would expect that its expected > output will need to be adjusted, unless it ignores this notification; > and if there isn't any test that attempts an export with externals, > then could you add one? The normal testsuite mechanisms ignore all the output lines affected by the patch. I suppose one could write a special test that checks the lines ignored by the usual test mechanism but any such test would have to work with all the output orderings that export could produce. That might be non-trivial. -- Philip Martin | Subversion Committer WANdisco // *Non-Stop Data*
