On 2018-06-05, 15:03 GMT, Nick Coghlan wrote:
> I was actually looking into this recently to see if the
> repository import feature could be used to run a regularly
> updated repository mirror that included all issues and PR
> comments in addition to the code,
Good, thank you very much. I didn’t, so I just worked out of the
PR materials and documentation on their side. I am glad somebody
did.
> I'm more confident in my ability to predict Microsoft's
> business incentives based on the prevailing technology
> landscape than I am in my ability to predict the actions of
> a VC firm like Andreesen Horowitz :)
Perhaps. I still would be more comfortable, if we were thinking
from time to time about alternatives in case Microsoft (or
somebody else) returns to The Bad Old Ways. I hope it won't
happen, but it might.
Backing up the git repo is not terribly troublesome because you just need to to it regularly by cron job. Plus since git is distributed we have copies of that repo all over the place and you can verify integrity through the commit hashes, so even without an official backup we have a ton of unofficial backups. :)
As for the PR data, that could be done by recording every webhook event from our repo if someone wanted to. Then you could reconstruct things by essentially replaying the log of events. There are also probably more structured ways to do it as well if people cared.
But the key thing is someone has to take the time and effort to set something up. I'm personally not planning to put in the effort since I think there's a massively bigger chance we will switch hosts again rather than GitHub deciding to shut off data access with no data export feature or lead time such that we have to craft our own solution and we can't do it fast enough to prevent the loss of data (and I don't think doing this ahead of time for an eventual migration is worth it either as any platform we move to might have its own migration support, etc.). IOW I ain't worried, but I think everyone assumed that for me. :)