bottle
bottle copied to clipboard
bottlepy
Is bottle dead?
I'm using bottle extensively in production but I'm a concerned by the last official version (0.12.13) release being almost 2 years old. Do you have plans to put all the developments from the dev release into the official (pip installable) release and also close the issues (and maybe include some new developments/improvements)? Thanks
Not dead, but I struggle to find enough time to finalize the 0.13 release. It grew quite big (again) and while it should be backwards compatible as promised, I am not confident enough to just release it and fix the bugs later. Bottle has a 'stable' release strategy and deprecation handling since 0.10
The current 'stable' release is 0.12, and security issues and critical bugs will be fixed in a timely fashion, as always.
@defnull I think you mentioned somewhere else that having people try switching their existing applications to 0.13 would help improve your confidence in 0.13. Is this still the case?
If so, it seems like it'd be helpful to advertise that a bit - maybe with a comment in the README, and a link to an issue for collecting feedback? Cutting an alpha release that people could more easily install might also help, with appropriate caveats around stability.
@defnull Hang in there brother. Bottle is one of the best web frameworks I have used. It will not die as long as it continues to be useful. The power of opensource will ensure that. I will try to contribute when I get enough time.
@defnull I have a little time to contribute, how could I help and what would be the main problems to solve. Thanks!
I just released 0.12.14 to get used to this again. The next bugfix release should be less work form me, as I updated my local scripts and workflow to the new tools (e.g. twine). This does not solve the 0.13 issue (too big) but it's a start.
Maybe It would be a good idea to add more maintainers to the project. There are a lot of PR open. I would like to contribute too, but without someone that have enough time to take care for contributions at least, it wouldn't be nice. For example, somethings that I could implement may be already done in some PR, duplicating efforts. And with a lot of PR open without proper integration, the code begin to diverge too much. Some cases like this is solved by a main fork that everybody follows that integrate the PR, may fix some known issues and keep the project alive. But I think this is something to use if there is no other solution. And It seems to me that just finding a new contributor with to carry the project for a while, at least the community contributions, would be enough.
One prominent issue is the high number of open PRs and issues. A lot of these can simply be closed, because they are outdated, dublicates or actually a support question and not a bug. I find it harsh to say 'No' to PRs or feature requests without a good explanation, and that takes time. I can definitely use some help here.
For that you need to increase the number of people with Permissions. I know that myself and @hartwork have started to contribute enthusiastically but got stopped because we didn't receive responses. You can add maintainers and give them gradually more permissions. New maintainers does not mean you automatically loose all control over the code base.
@defnull How does it happen that a missing feature of passing all server options to waitress is not available in the stable release 0.12 while it has been fixed in 2016? #826 0.12 code
I can understand that some featurees has been pushed to 0.13 but with the current release it's impossible to use multithreaded cherrypy (#975) and using waitress properly requires digging through really old issues while documentation presents an incorrect view. And by the way, docs are incorrect on both releases - config is not documented, 'options' does not exist and actually kargs are later passed to server.
I'd be happy to contribute a patch for 0.12 release if it makes way to pip package at some point.
If you want to back-port a bugfix to 0.12, go ahead. This is why these issues are still open. Pull requests are welcomed. 0.12.18 was released in December, so there is no reason to think new releases won't happen. As long as you do not break existing applications and the patch quality is good enough for a stable branch, even back-ported features have a good change get merged.
Also, it is far from impossible to use cherrypy, cheroot or any WSGI server you want. Server adapters are completely optional. Bottle works with any server that supports WSGI. Just follow their documentation.
I simply use version 0.13-dev. It would be nice to have a versioning of the development versions and be able to install them with pip without having to point a hash of git.
I prefer to deal with problems that have been generated recently than with old problems, the community is very active and I know that it will not take long to solve.
