[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#768797: HTTP method parser gets into wrong state upon 416 reply with Content-Length > 0



Hi,

no time at the moment to look at the issue itself, but:

On Sun, Nov 09, 2014 at 12:00:24PM +0100, Michael Stapelberg wrote:
> I would have tested and verified this and sent a patch, but there’s no
> testsuite for this part of the code, so I couldn’t easily verify that
> this was the problem and that my fix is correct, and neither could I
> verify that the suggested fix doesn’t break anything else. Hence I’ll
> leave it up to you to fix it.

You missed the scripts in test/integration/. They do all sorts of stuff
including starting our own webserver and running http/https against it.
You can run them all with ./test/integration/run-tests or a specific one
like ./test-partial-file-support which might be a good explicit test for
this part (just because I remember writing it for 416 testing).

As said we have our own simple webserver, so if you need to let it
behave in a certain (buggy/strange) way that should be relatively easy
to do as well.

These tests need some additional dependencies to work (and they take quite
a bit of time to run) so they aren't run in the build process, but they
are setup for autopkgtest in debian/tests/control and run by ci.d.n as
well as travis-ci.

Not the same as proper unit tests (not even close), but I am happy
enough that we got these in the last few years.


Best regards

David Kalnischkies

Attachment: signature.asc
Description: Digital signature


Reply to: