@flying-sheep This looks really good now BTW.

@flying-sheep This looks really good now BTW.

thanks! :D

@flying-sheep Don't invest too much time in this yet, @mitsuhiko must decide this ultimatively.

well, this is already discussed and implemented, everything that comes now is mere minutes of work, so i might as well.

otherwise i trust mitsuhiko’s judgment except in the matter of Python 2 vs Python 3 ;)

and regarding judgment: do you think the cleanness and neatness of grouping exception handlers using their http codes (or lack thereof) is worth the fact that it’s unnecessary for the logic? or not?

I don't think it's worth it, but @mitsuhiko may come up with a reason why it is absolutely necessary (I've encountered such surprises before).

can we now start the magic dance that summons him?

I still don't know how that dance works. But I usually spam him on IRC regularly.

@flying-sheep Would it be possible to fix #503 in the run?

Requesting help on reviewing this PR. I think it's a great change that definetly should land in 1.0.

btw. the regression test fails on master currently so i din’t break anything, i just rebased :)

btw. the regression test fails on master currently so i din’t break anything, i just rebased :)

It doesn't fail for me (I reverted the offending commit a while ago)

oh, strange. it doesn’t fail on travis. wat.

locally, test_memory_consumption fails.

so what did mitsuhiko say on IRC? you asking for reviews implies that he likes this?

Hmm, how does #1407 behave for you then?

so checkout unitaker/flask#no-threads and try? will do

/e: still test_regression.assert_no_leak fails

OK, so behavior-wise the current PR is excellent as-is.

The only thing missing now are docs. From a quick glance at errorhandling.rst there doesn't seem to be anything contradictory in there, but we should definetly clarify the current behavior.

We also need something in upgrading.rst, and a changelog entry.

writing something up right now, but errorhandling.rst doesn’t seem to contain anything about error handlers, only logging and debugging AFAICS

We need to document how Flask prioritizes errorhandlers, doesn't matter where. errorhandling seemed like the most related page to me.

Thanks @flying-sheep!

been a pleasure 😄