ecosystem-wg
ecosystem-wg copied to clipboard
rust-lang-nursery
Blitz more Libz!
The libz blitz offers a great framework for exploring, evaluating and polishing libraries. Should we do more evaluations for stabilising libraries this year? If so, what libraries do folks think could benefit from a blitz?
Should we do more evaluations for stabilising libraries this year?
Yes. Please note there was the wish for a better name last year.
Please note there was the wish for a better name last year.
Ah I wasn't aware of that, and can see why some folks might prefer a different name. What if we just call them 'evaluations'?
How about making https://github.com/Gilnaa/globwalk the next https://github.com/rust-lang-nursery/glob ? That would accomplish some of https://github.com/rust-lang-nursery/glob/issues/59 provided all the other issues can be fixed.
As I brought up in the CLI WG, what about helping to organize libz blitz centered around each of the other WGs. For example, the CLI WG could identify features we need, crates that fulfill those features, and then work on a libz blitz to help get them fleshed out.
I would like to bring chrono to this table, although I am not entirely sure how the group can intervene. https://github.com/chronotope/chrono/issues/195 Although the library is in a fairly working state right now, multiple breaking changes are desired before a stable release, and the current maintainer is lacking the time to fulfil these ideas. Considering also that there are over 500 direct dependents, diesel being one of them, it seems to me like a project worth catalysing.
@Enet4 chrono has been on my mind for a while. @quodlibetor might have some thoughts on where/how contributions could be the most helpful.
@epage Using the domain working groups as a source for libraries to focus on sounds good to me! Particularly libraries that are common between use-cases in multiple domains. Maybe we should start a thread on one of the forums to collect a set?
@Keats Should we suggest globwalk on https://github.com/rust-lang-nursery/glob/issues/59? We probably need to try revive some interest in that issue.
@skade I've opened #15 to work on a new name, so we can keep the threads a bit separate.
Yes, I've been thinking about improving chrono's API for a long time. I've been intending to write up my plans, as well, perhaps blitzing would give me the kick I need to actually do it? It'd be welcome, anyway.
@KodrAus the globwalk crate came from https://github.com/BurntSushi/ripgrep/pull/765 so I believe we just need to improve on it first to make sure it satisfies all the requirements of the current glob crate
Also (@KodrAus @Enet4) related to the state of chrono: one of the big issues that I had in the past was that the primary maintainer was essentially out of the picture and, although they owned the repo and crates.io, they weren't responsive or making releases. They've since transferred ownership to me, which gives me a lot more emotional wherewithal in terms of actually doing work for the project. It also means that my old philosophy of not seeking out contributions because it could be so disheartening to do work and have it not released for a year goes away.
Hopefully by this time next year we'll have a much larger group of people maintaining chrono, with at least one other releaser, and ideally a 1.0 release. There are big challenges to overcome to get there, but (other than multiple calendar systems, which I think might block 1.0 and which I don't know what they should look like) I think we can get pretty far.
Until we get an official ~blitz~ going for chrono I've written up some notes about some of the most egregious things I think need to be iterated upon: https://github.com/chronotope/chrono-rfcs/blob/master/notes/api-issues.md
I've spent a bit of time switching some types around and redesigning some traits and I think we can get to a much simpler API while still maintaining all of chrono's current power.
@quodlibetor @Dylan-DPC @Enet4 I thought we could move the discussion about chrono into #16.
@Keats Thanks for clarifying! I think I understand the background around glob, globset and globwalk better now. It doesn't look like there's much that needs to block deprecating the current glob implementation, but I left a ping on https://github.com/rust-lang-nursery/glob/issues/59 and opened #17 here.