On Thu, 16 Sept 2021 at 09:15, Richard Biener wrote: > > On Thu, Sep 16, 2021 at 1:37 AM Paul Koning wrote: > > Thanks, that's useful. Suppose I want to collaborate with one other person > > (for now) on pdp11 target work, does it make sense to keep that in a user > > branch since the community is so small and isolated? I assume the other > > person would need (as a minimum) write-after-approval privs. > > Note another option is to host int on github or gitlab and base it off > their mirrors of the GCC repo. > > Without write-after-approval access the person would not be able to > push to the repo > at gcc.gnu.org but of course you could do this with changes sent by > mail for example.
With Git it's really not necessary for everybody to write to the same repo. The other collaborator could have their own repo hosted on gitlab or github and send pull requests to you, for you to merge into your repo. Then you could push your branch (containing commits from you and the other collaborator) to a devel or user branch on gcc.gnu.org if you want (or not if you don't want to). Or you could both write to the same repo hosted on gitlab/github/wherever, by granting access appropriately. And then push the results to a branch on gcc.gnu.org (or not). With Subversion that didn't really work, there was no easy way to keep in sync with the gcc.gnu.org trunk unless your devel branch was also hosted on gcc.gnu.org, but with Git it's mostly irrelevant whether you host your branch on gcc.gnu.org or not. If you have no other way to host a repo that all collaborators can pull from (and you don't want to exchange patches by email) then hosting a branch at gcc.gnu.org is a good solution. But that's rarely the case, because there are numerous zero-cost repo hosting platforms available that most people can use (unless they are banned from all those platforms or firewalled in some way that makes them reachable).