On Sat, 2016-09-03 at 20:35, Waldir Pimenta <[email protected]> wrote: > One problem with forks (for the case where it can be assessed that the > author has indeed abandoned the project) is that existing issues, pull > requests, stars and watchers aren't transferred to the new repo; in fact, a > new repo is not even what's needed at all in this situation: new > maintainers is what's needed. > > I wonder if the Julia community should decide on some sort of guidelines > for handling these cases. For example, we could require any package > registered to METADATA.jl to provide administrative access to at least one > person from the core maintainer team (I mention admin access rather than > just commit rights, because then that person would be stuck as the > maintainer if the original author goes MIA, since they wouldn't be able to > give others --i.e. prospective new maintainers-- commit access). Of course, > the guidelines should also specify that this admin access should not be > used unless the author indeed goes MIA (say, for a set period of time > without warning). > > One nice side-effect of such a procedure would be that if a serious bug or > security issue is discovered in a dormant package, it can be addressed > without disabling the package altogether from METADATA.jl.
This sounds like a good idea (except that it makes more work for the core team). There has been talk of splitting METADATA into a curated one and a free-for-all (as it is now). Maybe this suggestion could be used just for the curated METADATA? > On Friday, September 2, 2016 at 11:48:13 PM UTC+1, Chris Rackauckas wrote: >> >> You can fork, update it, and then put a PR in to METADATA which changes >> the url for the package. Someone like @tkelman will probably try to contact >> the author to make sure he/she's really disappeared. Open an issue on >> METADATA and see if the author shows up. >> >> On Friday, September 2, 2016 at 3:34:42 PM UTC-7, Evan Fields wrote: >>> >>> I use the package GreatCircle <https://github.com/acrosby/GreatCircle.jl> >>> for >>> great circle distance calculations. On 0.4.x it generates depwarns, and >>> it's incompatible with 0.5. I've opened a pull request with the tiny >>> changes needed to use the package on 0.5, but there's been no response and >>> from the author's Github profile it looks like he/she is no longer >>> generally active. Is there a way to rescue the package in this situation? >>> (Besides perhaps forking and renaming, which leads to cluttered package >>> names...) >>> >>
