Phabricator guidance
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? I can't find any documentation of 'arc patch'. Can anyone else? It would be good to document this workflow, since it's a very useful one. Thanks Simon | -----Original Message----- | From: noreply@phabricator.haskell.org | [mailto:noreply@phabricator.haskell.org] | Sent: 06 October 2014 10:12 | To: Simon Peyton Jones | Subject: [Differential] [Commented On] D202: Injective type families | | jstolarek added a comment. | | Bump. Austin put this on feature list for GHC 7.10 and I'd really like | to this feature to make it into the next stable version of GHC. | Currently I'm not making *any* progress because of the "Not in scope" | errors. I've spent time looking at the code, seeing how things are | done in other places but none of this got me closer to solving the | problem :-/ Help? | | REPOSITORY | rGHC Glasgow Haskell Compiler | | REVISION DETAIL | https://phabricator.haskell.org/D202 | | REPLY HANDLER ACTIONS | Reply to comment, or !reject, !abandon, !reclaim, !resign, !rethink, | !unsubscribe. | | To: jstolarek, austin, simonpj | Cc: thomie, goldfire, simonmar, ezyang, carter
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...
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@gmail.com] | Sent: 07 October 2014 10:05 | To: Simon Peyton Jones | Cc: ghc-devs@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...
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@gmail.com] | Sent: 07 October 2014 10:05 | To: Simon Peyton Jones | Cc: ghc-devs@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@haskell.org http://www.haskell.org/mailman/listinfo/ghc-devs
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? S | -----Original Message----- | From: Jan Stolarek [mailto:jan.stolarek@p.lodz.pl] | Sent: 07 October 2014 13:06 | To: ghc-devs@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@gmail.com] | > | Sent: 07 October 2014 10:05 | > | To: Simon Peyton Jones | > | Cc: ghc-devs@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@haskell.org | > http://www.haskell.org/mailman/listinfo/ghc-devs |
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@p.lodz.pl] | Sent: 07 October 2014 13:06 | To: ghc-devs@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@gmail.com] | > | Sent: 07 October 2014 10:05 | > | To: Simon Peyton Jones | > | Cc: ghc-devs@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@haskell.org | > http://www.haskell.org/mailman/listinfo/ghc-devs |
_______________________________________________ ghc-devs mailing list ghc-devs@haskell.org http://www.haskell.org/mailman/listinfo/ghc-devs
participants (4)
-
Herbert Valerio Riedel -
Jan Stolarek -
Simon Marlow -
Simon Peyton Jones