UV Grant Report for 2020-12
Sat, 23-Jan-2021 by
Moritz Lenz
edit post
Paul "LeoNerd" has submitted his report for his work on the UV grant. It looks like the grant is close to finishing.
Due to technical problems, publishing of this report has been delayed, with no fault to Paul.
He write (with slight redactions due to formatting and bringing a conversation into a one-shot report):
---
Almost all the handle types are now done; except for `uv_async_t`.
I'm not really sure it makes sense (or is even possible) to wrap this from Perl, because it's inherently dealing with multi-threading issues which don't really come up. I suppose it (or at least the tests for it) could be optional on a (working) thready perl.
Almost all the request types are now done; except for any of the
filesystem ones:
Those should be relatively simple to write, if a little awkward to test some of them because they'll require filesystem mutations at testing time.
If we're not doing `uv_async_t`, then there's probably no need to wrap the thread pool functions, for similar reasons
There's a few small bits and pieces on UDP sockets that I skipped for now because they seem like they're not going to be used much, but I could probably do them for completeness: .
And finally, there's some more argument types that need handling for `uv_process_t`, except that libuv itself doesn't really document them clearly enough:
[I have an issue open with them about it](https://github.com/libuv/libuv/issues/3025).
Aside from those, I think we're basically done with code writing.
Code is currently sitting at .
---
Paul also published a [Trial release on CPAN](https://metacpan.org/release/PEVANS/UV-1.901-TRIAL), and later a [production release](https://metacpan.org/release/UV).
Paul concludes:
> I've also prodded the Mojo folks about it, so they can have a go at the Mojo::Reactor for it too. That all helps demonstrate the "Completeness Criteria" part of the original proposal.
Category:
(none)
Comments (0)