[GHC DevOps Group] State of CI

[Simon Peyton Jones]

Interestingly, I think this discussion has really teased out a key difference - perhaps the reason people prefer different tools here is because they're starting from different perspectives about what the tools are for?

I think you are right here.   Personally, I find that Phab focuses my attention on the *code*, rather than the design or architecture.  Even where discussions about the latter happen, they are buried in a sea of noise about low-level suggestions.   So my personal preference is to keep the higher level stuff on Trac (or even a wiki page) and regard the code review discussion as ephemeral.

Simon

From: Simon Marlow <marlowsd at gmail.com>
Sent: 06 June 2018 13:44
To: Simon Peyton Jones <simonpj at microsoft.com>
Cc: Manuel M T Chakravarty <manuel.chakravarty at tweag.io>; ghc-devops-group at haskell.org
Subject: Re: [GHC DevOps Group] State of CI

I think this reflects a different philosophy about code review. If we say that code-review is only for small-scale suggestions about the actual code changes, then yes I'd agree it's not all that useful to keep that history around. But we can (and sometimes do) have high-level architectural discussions as part of code review too, and indeed I think it's arguably the right thing to have these discussion around concrete code proposals. We keep all the discussion of the code in one place, and the discussion is attached to the evolving code patches.  In this world, the code discussions really are valuable history.

Interestingly, I think this discussion has really teased out a key difference - perhaps the reason people prefer different tools here is because they're starting from different perspectives about what the tools are for?  I don't think there's one true way here. Personally I've been exposed to multiple different working styles, especially having one foot in industry and one in the open-source world, and I've seen different philosophies that work.  We could as a project decide that we don't want to go the way of having high-level discussion alongside code review, in which case that's OK (I would slightly prefer to move in the other direction, but that's just my preference).

Cheers
Simon


On 6 June 2018 at 12:58, Simon Peyton Jones <simonpj at microsoft.com<mailto:simonpj at microsoft.com>> wrote:
We should keep in mind, though, is that past code reviews is valuable content that we can't discard,

Like Manuel I’m not at all sure about this.

I *do* regard the Trac conversation as a long-term asset, and often refer to tickets from Notes, as a way to say “here’s a more extensive discussion of what’s in the Note”.

But the Phab discussion of “please refactor this or that” seems far less valuable.  And actually I don’t think it is substantially cross-referenced from elsewhere.  Where there is substantive conversation about the approach, I’d rather see that on Trac.

So for me, long term access to the code-review trail is not very important

Simon

From: Ghc-devops-group <ghc-devops-group-bounces at haskell.org<mailto:ghc-devops-group-bounces at haskell.org>> On Behalf Of Manuel M T Chakravarty
Sent: 06 June 2018 12:50
To: Simon Marlow <marlowsd at gmail.com<mailto:marlowsd at gmail.com>>
Cc: ghc-devops-group at haskell.org<mailto:ghc-devops-group at haskell.org>
Subject: Re: [GHC DevOps Group] State of CI

Am 06.06.2018 um 19:11 schrieb Simon Marlow <marlowsd at gmail.com<mailto:marlowsd at gmail.com>>:

* None of this work is GitHub specific. Nor all that CircleCI or
Appveyor specific for that matter (work is currently focused on
improving the test suite).
* Our GitHub lock-in factor is currently low to pretty much absent,
and would remain low even if the review workflow becomes more
systematically GitHub centric (it already is for some small
contributions).
* That's because tickets remain on Trac, and the code along with the
entirety of its history remains in a standard Git repository, GitHub
or not. Also because GitHub is not a CI provider, those providers we
do use integrate with other code hosting solutions (e.g. Appveyor with
GitLab), and the surface area of CI provider-specific code is small.

We should keep in mind, though, is that past code reviews is valuable content that we can't discard, nor can we easily migrate it to a different code review platform. At this point we have nearly 5K diffs on Phabricator, many of which have non-trivial code-review trails, and these are cross-referenced from Trac, emails, and other places. Even if we moved to github, we would want to keep Phabricator running so that we have access to this content, and people will experience friction though havng to deal with another system.

To me, the friction caused by the transition and the inability to do a clean move is more worrying than the missing code review functionality on github.

Actually, to me this is a red flag. Core reviews shouldn’t be essential documentation. I wonder what has been going wrong so that we have got to this situation.

If important points are uncovered or documented in code reviews, that information should be included in either the source code or associated documentation. (Having git history as an important record for the evolution of the code is fine as it is now standard to not only have one snapshot of the source, but a source control history of the source as the definitive record of a piece of software. And as we painfully discover, much in contrast to code reviews, the source control history is recorded in a widely understood standard format.)

Cheers,
Manuel


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.haskell.org/pipermail/ghc-devops-group/attachments/20180606/ead34133/attachment-0001.html>