[llvm-dev] [RFC] One or many git repositories?

Renato Golin via llvm-dev llvm-dev at lists.llvm.org
Fri Jul 22 02:48:56 PDT 2016

On 22 July 2016 at 00:51, Mehdi Amini <mehdi.amini at apple.com> wrote:
> While I see some good principled way to have a submodule umbrella repo in
> git, I don’t see any *without  server-side hooks* that does not have any
> flaw*. Unfortunately this is not addressed by Renato’s proposal, and github
> does not allow server-side hooks, and another git hosting service is spelled
> out-of-discussion for Renato’s proposal.

*Please*, stop calling it *my* proposal. It absolutely wasn't.

I'll repeat, as people seem to prefer repeated arguments than
references to past emails, but:

* I have sent a number of concerns and options
* People have favoured GitHub with sub-modules (I hadn't)
* I summarised the first proposal, which seemed to be reaching consensus

Let's call it "First Proposal" or "GitHubSubMod" proposal:


Everything "out-of-discussion" on the first proposal was in the
interest of reaching a self-contained proposal, and had absolutely no
ulterior motive.

Now the proposal is there, best we could make it. If there are
technical flaws, by all means, send a review to that document, but you
can't change that proposal into something else.

You can, however, create a new one, and that's what you're doing.

As people said earlier, getting to know one proposal well, has shown
many people that the "consensus" might not have been the best way
forward, but that was only possible by actually finalising at least
one proposal.

My assumption was that a survey would take us to the next step
(finding the precise and impersonal problems with that proposal), but
it seems I didn't need that. I stand corrected.

One thing your proposal doesn't even touch is where the repo will be.
I know it's basically orthogonal, but it's one of the key reasons why
we need to move. I have no preference, as long as the solution is
maintainable and cater for our needs.

My personal opinion is to host somewhere professional unless there's a
good reason not to.

If we use external hosting, GitHub is the best because there are
already thousands of forks (see Chisnall's email) there already, and
people do come to the list thinking the GitHub repo is our official

If we don't, we'll have to understand the costs and who's going to
maintain it (volunteer vs. hired help). Relying on volunteers (like
myself) is extremely risky and I'd very much rather not go that way.
Relying on any company can create bias (or the impression of bias),
which can divide the community.

Again, I'm not pushing *any* agenda, just laying out the issues. But
if you want to compete with the first proposal, you *have* to have a
complete proposal, with all the pros and cons clearly laid out.


PS: We may need a grid of proposals ({external, local} x {submod,

More information about the llvm-dev mailing list