Dylan Baker
I ran into this also (on 18.1), and discovered that oddly enough defining `__slots__` myself and using `attr.s(these={...})` does work. simplified example: ```python @attr.s(these={'foo': attr.ib()}) class Foo(Generic[T]): __slots__ = ['foo']...
This does work with python 3.7, at least with `__future__.annotations`, I haven't tested it without.
I applied this patch and can confirm that this fixes #1629 . Is there any chance of this landing and being added to a release?
Yes, as is stackedit tries to create a list of 9999 items (IIRC), but gitlab well silently cap that. With this patch (which is not in master), it parses additional...
@lorenx, sorry, I do development on github and get a *lot* of messages. I've been using dillinger.io, but github intergration is broken there too now, so I've just been using...
I can't reproduce the encode error, what I can reproduce is that the CC will show up as a quoted printable string, `=?utf-8?q?=C3=96we?= `, which is obviously not ideal. That...
wait, it's the address with the non-ascii in it? ohhhh, yeah, the python builtin definitely doesn't handle that.
I think @dkg was the one working on this?
That set of tests is fixed in the next commit, which seems to be for #1137, so what's supposed to be tested is that the subprocess.Popen call is invoked with...
I'm marking this as RFC, since I'm thinking about other things we could do that might be better.