Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

#1 can be done by creating a separate submodule repo that only stores docs+issues files. It's up to the repo's users to agree on a system by which the files should be organized, but it's doable.

I'd propose directories "issues/open", "issues/closed", with each issue filename being "{created:yyyy-MM-dd} - {subject}.md". Symlinks could be used to track ownership/responsibility if each repo contributor has their own directory in the repo too.



That's an awful lot of extra work to insure against maybe 1 hour per year of downtime.


Just because present state is 99.7579% uptime (for the month) doesn't mean it will always be so.

You back up your data, why shouldn't you backup your github data?


Backup is one thing, choosing to run a crappy manual system just in case a vendor goes down is entirely different.


It doesn't have to be a "crappy manual system" - I'm simply suggesting that given that git itself is a damned good distributed versioning database for arbitrary content, then we might as well also use it for distributed issue-tracking. A simple offline-mode browser-based editor that lives in a single HTML file within the repo would provide a nice GUI on top.

Hmm, I think I might be on to something... anyone want to start a project?


>git itself is a damned good distributed versioning database for arbitrary content, then we might as well also use it for distributed issue-tracking.

For what it's worth, it's interesting to see that the Fossil distributed SCM includes an issue tracker but they made a deliberate architecture decision to not propagate the tickets data.[1] They had a chance to make your "distributed-issues-tracking" idea a 1st-class concept in Fossil but decided against it.

Also, the issues/tickets is just one example feature. Github will continue to evolve to add more and more sophisticatted SDL/ALM (application lifecycle management) like JIRA and Microsoft Team Foundation Server. Those features are not easy to implement in a peer-2-peer SCM with practical usability.

[1] https://fossil-scm.org/xfer/doc/trunk/www/qandc.wiki


Thank you for the link. I read through their justifications and I think using a git-submodule solves their problems of polluting the main project history and permissions issue. Using directories for mutually-exclusive state grouping (e.g. "closed"/"open"/"new") solves the directory problem.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: