liblinear
liblinear copied to clipboard
cjlin1
Pull requests not acted upon
If not, would there be interest in maintaining a community fork with some people?
Sorry not at this moment. Having a group of people to change certain things such as interfaces/makefile or ways for installation is good.. However, we are very careful about the change of algorithms. We feel that any alg change must be made only after extensive analysis and evaluation. wvengen writes:
If not, would there be interest in maintaining a community fork with some people?
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.*
@cjlin1 thank you for your reply! I think it's a very good idea to thoroughly test any changes touching the algorithms (even when 'just' parallelizing). Thank you for guarding that.
Given community interest (which I'm probing here as well), I'm hoping there may be some way to get community contributions integrated. I think there are some good improvements in the queue (#14, #15, #18, #28, #29, #30 and probably more), and until they are merged they won't see the light of day in distribution packages, nor will many users will be able to reap those benefits until then.
So my first question would be to @cjlin1: assuming that you see value in improvements by others than yourself and the machine learning group, and assuming that you'd like to guard proper working of the algorithms (and I, for one, very much welcome that), and assuming that you might not have that much time for maintaining this, is there anything we can do to have both?
Some ideas: a team of PR reviewers; a team of community maintainers that assigns issue labels and verifies whether changes are in the build system, platform support, usability, or the algorithms; or starting a semi-official community fork where maintainers agree to only include non-algorithmic changes, which you could merge at your convenience.
If you're not really in for this, perfectly fine, I understand that you have research to attend to. But if you'd welcome community involvement, I'd be happy to see if something can be worked out.
In any case, I'd love to hear if there are other users (and developers) who would be interested in getting this library forward - don't hesitate to :+1: or comment.
I submitted my pull request more than two years ago, and you only responded to it four days ago. I have since moved on to other projects. Sorry!
@cjlin1 It is disappointing that these pull requests are not being addressed. Explicitly rejecting the pull requests would be more responsive.
If pull requests are not going to be considered, then perhaps you should add a comment to that effect so that other developers to not expend time trying to improve liblinear.
I think that we should create a community fork. I am willing to be one of the organizers. I would want to add a number of regression tests and unit tests so that we can be more assured that the requests are reliable.
If we create a community fork, we will want to move over the issues as well.
It seems this may be the only option to develop a community the benefits from each other.
The other option is to have @cjlin1 make one of us an administrator of this repo.
If @cjlin1 is willing to do that, it would be the most effective and most conducive to supporting the community here. I would recommend the @simsong is made an administrator.
@simsong I made a fork and a refactoring of the project introducing CMake as build system to compile the project easily on Windows, and I introduced Openmp to parallelize some routine. I hope it is helpful, the fork is: https://github.com/angleto/liblinear
All open issues have been migrated. @angleto , will you migrate the pull requests?
Excuse me @simsong, where should I migrate the pull request: which repository/branch?
To https://github.com/angleto/liblinear
Sent from my phone.
On Jan 23, 2018, at 4:13 AM, Angelo Leto [email protected] wrote:
Excuse me @simsong, where should I migrate the pull request: which repository/branch?
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.