[Reviewers] Recommending Phabricator

[Yuya Nishihara]

On Fri, 9 Feb 2018 17:53:43 +0000, Ryan McElroy wrote:
> Indeed, and I bought a contract with the Phacility company for a few 
> months after the last sprint to talk about some possible changes. I got 
> a special deal while I tried to convince FB to pay for the contract -- 
> but since Evan Preistley (main guy at Phacility) wasn't into the 
> directions we wanted to go, he only offered to produce one-off patches 
> that wouldn't be maintained for many of the ideas I solicited from the 
> community after that sprint.
> 
> >
> >> On Feb 8, 2018, at 13:03, Yuya Nishihara <yuya at tcha.org> wrote:
> >>
> >> On Wed, 7 Feb 2018 09:54:39 -0800, Gregory Szorc wrote:
> >>> One of the changes I made was to recommend Phabricator over emailing
> >>> patches. I did this because I think Phabricator is more friendly to new
> >>> contributors and provides a gateway for new contributors getting more
> >>> involved.
> 
> I'm +1 on more Phabricator, but I'm not reviewing much right now so you 
> should ignore my opinion on this.
> 
> >> I hope not advertising Phabricator more until its email interface gets fixed.
> >> Proper threading and inline-reply are must IMO.
> 
> The main reason Evan pushed back on this is that he views emails from 
> phabricator as "notifications" more than "mediums for consumption". He 
> thinks that the main interface for phabricator should remain the 
> website, and that it should not be a tool for review-over-email. He 
> suggested that he could be paid to write a one-off patch that would 
> produce emails the way we want, but that future changes might (and 
> likely would) break this patch, and we'd have to maintain it, or pay for 
> an updated patch.

First, many thanks for the details how things was going.

> >>> In addition, I think the time has come for more of Mercurial's core
> >>> contributors to transition from email to Phabricator.
> >> I thought that was a temporary issue of email infrastructure at Google?
> 
> I'm not sure what Google's email infra has to do with this; I think Greg 
> is suggesting that more core hg contributors (including everyone in this 
> list) start submitting their patches to phabricator.

I meant I though Googlers started using Phabricator because they couldn't send
patches to the list. Well, I'm hoping they'll be back to the list.

> >>> As a reviewer, I've noticed I'm biasing to looking at Phabricator first
> >>> then falling back to Patchwork if I still want to do reviews. Phabricator
> >>> is just a more pleasant experience for me as both a reviewer and a
> >>> contributor.
> >> I do truly the opposite. Ignoring almost all Phabricator emails, and pick
> >> random patches, which I would have to review, from Yadda using phabread.
> 
> This may just be one the the "great debates" like emacs vs vim, that 
> will never be fully resolved. I certainly have found it far easier to 
> deal with patches from phabricator than I did dealing with email 
> patches, but I have years of experience with a phabricator-like workflow.
> 
> The key difference for me is that email might be slightly better if I 
> plan to read every patch. It's almost impossible, without more advanced 
> tooling, to effectively ignore future versions of a patch I'm not 
> interested in (because of v2, v3, etc).
> 
> However, during the times I have spent less time reviewing (like 
> currently), phabricator is great for me to do "drive-by" reviews, 
> occasional spurts of reviewing, etc. All the context is collected into a 
> single place for the given patch; I don't have to search my inbox or set 
> up tooling to make it work well.

Yeah, perhaps it's great tool for reviewing a handful of patches you have
expertise.

I want to process patches including ones I have near-zero knowledge nor
interest, in a limited time (<1hr per day, hopefully.) Phabricator doesn't
help and I'm lazy to remember different key bindings of web toolings or to
make them acts like Emacs. Regarding the email UI, its hugely decreased S/N
ratio matters for me.