Deal with fixture init failures in boost.test#31
Deal with fixture init failures in boost.test#31nicoddemus merged 2 commits intopytest-dev:masterfrom
Conversation
On sufficiently fatal errors (such as an exception thrown from a global fixture ctor) boost.test ignores --output_format=XML and dumps the error message as is. Check for those cases explicitly instead of crashing due to malformed/empty xml. (alternatively we could check for result code 200 (aka boost::exit_exception_failure), but I'm not sure that boost.test doesn't return it in other cases)
|
Thanks a lot for the PR, appreciate it! |
|
Just released |
|
Wait, it looks like you forgot to push 0.4.4 changes to github actually, there's no version bump nor tag. By the way, how do you make the archives you publish on PyPI? It would be nice to have .tar.gz back, especially if the files inside weren't checked out with Windows line endings (not that it causes any real problems, but still)! Also, it's entirely irrelevant, but while I'm at it: newer versions of boost.test offer a |
3 participants
On sufficiently fatal errors (such as an exception thrown from a global fixture ctor) boost.test
ignores --output_format=XML and dumps the error message as is. I added an explicit check for those cases so the plugin no longer throws a weird exception from the depths of lxml due to malformed/empty xml.
(Alternatively I could check for result code 200 (aka boost::exit_exception_failure), but I'm not entirely sure that boost.test doesn't return it in some other cases, and I think that this approach is good enough)