Phabricator guidance

[Simon Marlow]

On 07/10/2014 15:52, Simon Peyton Jones wrote:
> Much better thank you.
>
> Now I get as far as
> compiler/typecheck/TcTyClsDecls.lhs:421:26: Warning:
>      Pattern match(es) are non-exhaustive
>      In a case alternative: Patterns not matched: KindedTyVarSig (L _ _)
>
> No time to investigate right now.
>
> =============
>
> New workflow question for Austin/Herbert.  Suppose Jan updates D202 to fix the above.  What's my workflow for getting his fix. If I say "arc patch D202" again, it probably won't work well because I'm already on local branch arcpatch-D202.
>
> I suppose I can
> 	git checkout master
> 	git branch --delete arcpatch-D202
> 	arc patch D202
>
> but is there a better way?

"arc patch" always makes a new branch, unless you ask it not to.  So it 
doesn't matter what branch you're currently on.

Cheers,
Simon



>
> S
>
> |  -----Original Message-----
> |  From: Jan Stolarek [mailto:jan.stolarek at p.lodz.pl]
> |  Sent: 07 October 2014 13:06
> |  To: ghc-devs at haskell.org
> |  Cc: Simon Peyton Jones; Herbert Valerio Riedel
> |  Subject: Re: Phabricator guidance
> |
> |  Ugh. Arc is not easy to use :-/ Indeed 3e17822 does not seem to be in
> |  the revision on phab, although it exists in my local tree. I just
> |  pushed a fixed to Phab. Simon, does `arc patch D202` work now?
> |
> |  Herbert, remember how I complained on IRC that `arc diff` does not
> |  automatically recognize that I'm updating a revision and I need to
> |  manually specify base commit? You told me that I need to add revision
> |  information to the commit message. I did that and `arc diff`indeed
> |  recognized the revision without me explicitly specifying the base
> |  commit. But now it turns out that it created an incomplete revision by
> |  pushing only the latest commit from my branch :-/
> |
> |  Janek
> |
> |  Dnia wtorek, 7 października 2014, Simon Peyton Jones napisał:
> |  > Aha, that helps.  And looking further at
> |  > https://phabricator.haskell.org/D202, I can see under "Revision
> |  update
> |  > history" that there are four diffs all stashed in this on Phab
> |  ticket.
> |  > (That contradicts my previous model which was one patch per Phab
> |  > ticket; people have been complaining about that.)
> |  >
> |  > So my new questions are:
> |  >
> |  > * How can I apply "Diff 1" or "Diff 2"?  Using "arc patch" only
> |  > applies "Diff 4"
> |  >
> |  > * How can I apply all of "Diff 1" ... "Diff 4" in one go?
> |  >
> |  > Simon
> |  >
> |  > |  -----Original Message-----
> |  > |  From: Herbert Valerio Riedel [mailto:hvriedel at gmail.com]
> |  > |  Sent: 07 October 2014 10:05
> |  > |  To: Simon Peyton Jones
> |  > |  Cc: ghc-devs at haskell.org
> |  > |  Subject: Re: Phabricator guidance
> |  > |
> |  > |  On 2014-10-07 at 10:57:00 +0200, Simon Peyton Jones wrote:
> |  > |  > I suppose I will have to look at this. But I have no clue how to
> |  > | do
> |  > |
> |  > |  so.
> |  > |
> |  > |  > D202 itself seems to be a very small patch (only ten lines or
> |  > | so),
> |  > |
> |  > |  so presumably it applies on top of some other patch?  But what?
> |  > |
> |  > |  > Someone said I could use
> |  > |  > 	arc patch D202
> |  > |  > to apply the patch in my own tree, which is crucial for
> |  > | reproducing  > the error that Jan is stuck on.
> |  > |  >
> |  > |  > BUT the patch presumably applies to a  > particular commit, NOT
> |  > | the head of my current tree.  But what is the  > base commit to
> |  > | which it applies?  Does arc patch check out the base  > commit
> |  > | before applying?
> |  > |
> |  > |  If you actually perform 'arc patch D202', this is the output you
> |  > | currently get:
> |  > |
> |  > |
> |  > |  ,----
> |  > |
> |  > |  | Created and checked out branch arcpatch-D202.
> |  > |  |
> |  > |  |
> |  > |  |     This diff is against commit
> |  > |
> |  > |  3e17822f5f4e4d2f582dc0a053f532125f9777c7, but
> |  > |
> |  > |  |     the commit is nowhere in the working copy. Try to apply it
> |  > |
> |  > |  against the
> |  > |
> |  > |  |     current working copy state?
> |  > |
> |  > |  (3549c952b535803270872adaf87262f2df0295a4)
> |  > |
> |  > |  |     [Y/n] n
> |  > |
> |  > |  `----
> |  > |
> |  > |  So yes, 'arc' tries apply the code-revision on top of the commit
> |  is
> |  > | was based on; and in this case, it is actually missing from ghc.git
> |  > | :-  /
> |  > |
> |  > |  What's more, you can also declare that a code-revisions builds on
> |  > | top  of another code-revision, in which case 'arc' will
> |  > | automatically try  to
> |  > |  (recursively) apply that other code-revision to your source-tree
> |  > | first, before applying the one you are actually requesting on top.
> |  > |
> |  > |
> |  > |  I hope Austin or someone else may chime in to provide further
> |  > | assistance if this doesn't help...
> |  >
> |  > _______________________________________________
> |  > ghc-devs mailing list
> |  > ghc-devs at haskell.org
> |  > http://www.haskell.org/mailman/listinfo/ghc-devs
> |
>
> _______________________________________________
> ghc-devs mailing list
> ghc-devs at haskell.org
> http://www.haskell.org/mailman/listinfo/ghc-devs
>