[llvm-dev] Enable Contributions Through Pull-request For LLVM

Don Hinton via llvm-dev llvm-dev at lists.llvm.org
Tue Nov 12 10:53:20 PST 2019

-1 -- switch to PR's
+1 -- Hal's compromise proposal (as long as I can continue to use Phab)

I agree that the documentation could be better, but I don't see that as a
justification for switching from a superior tool to an inferior one.  Let's
work on the documentation first, then if there's still a compelling reason
to switch, do it then,

I share most of the concerns already mentioned, but my biggest problem on a
daily basis is the use of force pushes to PR's, which loose information (no
way to tell what changed from the previous version -- I deal with this
everyday, and have to re-review the entire patch because I can't see what
was changed/fixed) . Since the branch is in someone's personal fork, I
don't see how it could be outlawed, and that's a problem and huge time sync.

However, I do agree that the patches should be squashed into a single
commit when merged to master.

Btw, I've read several posts claiming github is willing to work with us to
improve the process.  Perhaps we could get them to adopt Phab.


On Tue, Nov 12, 2019 at 9:52 AM Robinson, Paul via llvm-dev <
llvm-dev at lists.llvm.org> wrote:

> In general I mildly favor moving to GitHub PRs, mainly because it is
> a de-facto standard in the industry and many people are already
> familiar with it.  LLVM's idiosyncratic and essentially undocumented
> tool is not, whatever minor advantages it might have in certain
> technical respects.
> David Tellenbach wrote:
> > One thing I always liked at the Phabricator workflow is that it is
> > based on diffs only.
> And other people on the thread find this an annoyance.
> Being unable to chain related reviews is about the only case where
> GitHub would appear to be at a significant disadvantage, and offhand
> it would not seem like an insurmountable technical obstacle if we can
> persuade the GitHub folks that it's actually pretty useful.
> Alex Brachet-Mialot wrote:
> > Just because more people might be familiar with GitHub, doesn't mean
> > it is superior.
> "The perfect is the enemy of the good."  Automobile controls are
> largely standardized; you might design a superior set of controls,
> but the simple fact of being different from all the others would be
> a barrier to adoption.  Even if GitHub's workflow is flawed, it is
> what new contributors will be used to, and new contributors are the
> lifeblood of the project. And as a major project hosted on GitHub,
> it is not impossible that we could influence GitHub to smooth it out
> to better suit our use-cases.
> Or, check out Hal Finkel's suggestion, which would allow casual
> contributors to at least start out with their accustomed workflow
> and then be funneled into LLVM's idiosyncratic review tool, which
> is fundamentally user-hostile but our community has gotten used to.
> > I think many people agree that Phabricator is really good,
> And many people agree that Phabricator is a major pain.  It has
> zero useful documentation (other than what the LLVM project has
> written for its own use) and IME makes every effort to make useful
> features difficult to find and use.  Ingenious people still manage
> to find and use them, but Phab absolutely does not make it easy.
> > Getting reviewers [on GitHub] is a much bigger barrier
> This would depend on how new reviews are advertised.  With Phab, we
> understand how they are advertised (i.e., mailing list).  With GitHub
> this is apparently not automatic, but still automatable, IIUC.
> --paulr
> _______________________________________________
> LLVM Developers mailing list
> llvm-dev at lists.llvm.org
> https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20191112/0246abe5/attachment.html>

More information about the llvm-dev mailing list