0.10.1 on PyPI #267
Comments
|
Bump, I am waiting to do a PyPI release on my project because I can't have git dependencies, but I need 0.10.1 in my project. |
|
Hello, are there any updates on this? Best wishes. |
|
Greetings, Best wishes and be healthy :) |
|
Any update on this? If not I'm going to just copy and paste the code from 10.1 into my repo. |
|
They are not paid to do this so your comment is unnecessary. This repo has an MIT license so this gives you a lot of freedom. |
|
If being paid would help, there's a modest amount up for grabs via Tidelift: |
|
@uiri bump/ping |
|
yay! looks like a new release is up on pypy |
|
woot woot! |
|
Hmm... there hasn't been much activity since 0.10.1. :( @uiri My offer from #267 (comment) still stands. |
|
@uiri It is rather sad and worrying that what seems to be the most popular TOML library currently has been entirely unmaintained and does not support the current, stable 1.0 version of the spec, despite a large amount of community interest in contributing and moving things forward. There seem to be several options, from most to least preferred: Ideally, this repository could be moved to a Github org with the appropriate contributors added, which would reduce avoid a single point of failure with it depending on a user's personal account; however, it would require the project adopt a new name on PyPI, and downstream users manually update their dependencies. Collaborator access could be granted to one or more experienced contributors, and development continued here. Finally, this repo could be archived and users pointed toward an active fork or replacement project. All of these, however, are better for everyone involved than simply keeping the status quo, as eventually either users will have to fork the project, choose a new name on PyPI and rebuild the community anew, or abandon it altogether and have it die a slow, easily avoidable death. It seems the whole community wants to support you, @uiri , and your amazing work here, and it would always be better whatever they end up choosing, to do so with your blessing. Given @pradyunsg , an experienced, well-known, clearly deeply subject-matter-knowledgeable community member willing to take on the work has offered to step in to move things forward, it would certainly be ideal if maintainership (and more importantly, PyPI deploy access) could be shared with them. However, after a year of no response, and very little Github activity by the current maintainer, it might be time to consider a fork, unless the community prefers adopting an alternative toml parser (Tomli seems nice, but doesn't have widespread adoption yet, and I've ran into a number of edge-case bugs and subtle incompatibilities with tomlkit for use cases that don't need to preserve style). |
https://toml.io/en/ uiri/toml#267 uiri/toml#279 (comment) While the TOML spec is alive and well, the `toml` package that was used in this project is dead (uiri/toml#267). There are several alternatives. `rtoml` is the fastest, but does not provide "round trip" guarantees (meaning that it can't load, then dump, and get an identical result). However, for simple loading and parsing it's fine. `tomli` is another alternative, but it is read-only (writing requires a separate `tomli-w` package), and requires files to be opened in binary mode to parse TOML. `tomlkit` is used by Poetry (from the Poetry author), but is currently 70x slower than `rtoml`. It's also important to note that the TOML parsing and settings loading is somewhat tangential to the focus of the inboard project. A separate project, https://github.com/br3ndonland/fastenv, is more focused on settings management, and may feature TOML support in the future. This commit will remove `toml` from the `fastapi` Poetry extras group. Note that `toml` is still a sub-dependency of some `dev-dependencies`, including pre-commit and pytest.
Good morning,
are you planning to update toml on PyPI too soon? It is now still at version 0.10.0.
Thank you,
-Filippo
The text was updated successfully, but these errors were encountered: