release timing
Alan & Kim Zimmerman
alan.zimm at gmail.com
Tue May 5 19:23:52 UTC 2015
I aplogise for the following long post on Api Annotations.
They are a new feature, and allow a whole class of tools to become much
simpler. The interest in this technology can be seen in the strong support
Matt Pickering's proposal for GSOC achieved, and that it was accepted. This
project is a good showcase for the annotations, as it should allow the
hints from HLint to be directly applied to the source code.
This technology can only be useful when it is generally available, which
means being baked in to the compiler. We came really close to getting it
all together for 7.10.1, but unfortunately some issues slipped through.
These were picked up when Matt Pickering ran the ghc-exactrpint utliity
over most of hackage, and found the remaining ones. The fixes for these are
now all queued up in trac/phab with patches. If they do not go in to 7.10.2
it means waiting for 7.12 to be in general use before the tooling built
from this can be used widely.
The patches themselves do not change the operation of the compiler, merely
re-arrange things slightly in the parsing stage, to make sure all the
annotations make it through to the final ParsedSource. The two most
intrusive ones are
D840, which introduces new parser productions to correct a pre-existing
error where `'[]` was parsed as a `HsTyVar` rather than a `HsExplicitListTy`.
These have different annotations so it is a problem; and
D836, which preserves the parsed `HsForAllTy` structure, including any
nested `HsParTy` wrappers. This results in the original flattening code
being moved out of `RdrHsSyn` into the renamer and typechecker, but does
not change any of the operations.
I know this places a huge burden on Austin in particular, because it
requires a lot of merge operations from the number of patches. The only way
I can propose to ease this is to submit a mega-patch with all the changes,
which can be applied at once, once the individual ones are reviewed and
found acceptable (with whatever fixes are required).
The full list of Trac tickers / Phab patches follows below.
Regards
Alan
---------------------------------------------------------
10207 - parser: ParStmt has incorrect SrcSpan
D803 (landed 7.10.2)
10214 - parser: TransStmt has incorrect SrcSpan
D806 (landed 7.10.2)
---
10209 - parser: opt_kind_sig has incorrect SrcSpan
D813 (landed master, scheduled 7.10.2)
10255 - API Annotations : ExprWithTySig processing discards annotated spans
D823 (landed master, scheduled 7.10.2)
---
10357 - ApiAnnotations : pquals production adds AnnVbar in the wrong place
D869
10358 - ApiAnnotations : PatBind gives wrong SrcSpan for the pattern.
D873
10254 - parser : the API annotation on opt_sig is being discarded
D822 (landed master)
10256 - parser: API Annotations : guardquals1 does not annotate commas
properly
D818 (landed master)
10268 - ApiAnnotations : quoted type variables missing leading quote
D825. (Accepted Austin) Depends D840/#10299
10269 - ApiAnnotations : RdrHsSyn.isFunLhs discards parentheses
D832 (Accepted Austin)
Note: Potentially keep HsPar as an alternate?
10277 - ApiAnnotations : lexer discards comment close in nested comment
D829 (landed master)
10280 - ApiAnnotations : AnnComma missing in TupleSection
D834 (Accepted Austin)
10287 - ApiAnnotations : BooleanFormula construction discards original
D837. Partial fix. Full fix by locating BooleanFormula properly.
10307 - Api Annotations: RdrHsSyn.mkAtDefault causes annotations to be
disconnected
D842
10309 - ApiAnnotations : mkGadtDecl discards annotations for HsFunTy
D848
10312 - ApiAnnotations: misplaced AnnComma for squals production
D846 (accepted Austin)
10299 - D840: Correct parsing of lifted empty list constructor
D840
--- The next four are all addressed by D836
10315 - ApiAnnotations : Empty context loses annotations
D855 retired, fixed by 10354/D836
10354 - ApiAnnotations : parens around a context with wildcard loses
annotations
D868 - no, superseded by D836
10363 - ApiAnnotations : HsForAllTy discards parens
D836
10278 - ApiAnnotations : Nested forall loses forall annotation
D836 (old/alternate D833)
---------------------------------------------------------
On Tue, May 5, 2015 at 1:16 PM, Simon Peyton Jones <simonpj at microsoft.com>
wrote:
> Just to be clear, the issue is this
>
> * For whom is 7.10.2 mission-critical?
> * And with what patches? (give specific tickets)
> * And by what date?
>
> The only reason to make a release is because someone is stalled, or
> seriously inconvenienced, unit it happens. Otherwise we should wait. So:
> "+1" doesn't give enough information. Of course, the sooner the better,
> the more bugs fixed the better.
>
> You might say "I'm waiting for API Annotations; but I'm under no time
> pressure". Or "My whole business is stalled pending a fix to Trac #xxx" or
> whatever.
>
> The whole API annotations thing is a special case. We usually make NO api
> changes in a patch-level release. But the API annotations stuff is brand
> new, and apparently most of the fixes depend on API changes. So I think we
> are going to let them sneak in. Do we have a comprehensive list, in Trac
> tickets of all the patches that are sought?
>
> The canonical list is
> https://ghc.haskell.org/trac/ghc/wiki/Status/GHC-7.10.2
> Only patches that are listed as "merge" or "patch" are even being
> considered. If you have something you care about that is not in that
> state, can you cause it to be in that state?
>
> Thanks
>
> Simon
>
> | -----Original Message-----
> | From: ghc-devs [mailto:ghc-devs-bounces at haskell.org] On Behalf Of
> | Matthew Pickering
> | Sent: 05 May 2015 12:09
> | To: Alan & Kim Zimmerman
> | Cc: ghc-devs at haskell.org
> | Subject: Re: release timing
> |
> | +1 for the API Annotations patches.
> |
> | On Tue, May 5, 2015 at 11:26 AM, Alan & Kim Zimmerman
> | <alan.zimm at gmail.com> wrote:
> | > I'd love to see the various queued API Annotations diffs go in.
> | >
> | > Alan
> | >
> | > On Tue, May 5, 2015 at 12:16 PM, Roman Cheplyaka <roma at ro-che.info>
> | wrote:
> | >>
> | >> I'd love to see GHC 7.10.2 once this haddock issue is fixed:
> | >> https://github.com/haskell/haddock/issues/385
> | >>
> | >> On 05/05/15 13:08, Simon Peyton Jones wrote:
> | >> > This time 'round I'd like to do a HP 7.10.2 concurrent with GHC
> | 7.10.2!
> | >> > Crazy, I know, and who knows if we can pull it off....
> | >> >
> | >> >
> | >> >
> | >> > Good plan!
> | >> >
> | >> >
> | >> >
> | >> > In Austin’s last GHC Weekly News we asked if anyone had any
> | >> > constraints on the release of GHC 7.10.2. So far as I know, no
> | one
> | >> > replied. *So the current status is that we’ll hold it until
> | >> > someone says “getting
> | >> > 7.10.2 out really matters to me”.* Other things being equal, the
> | >> > longer we wait, the more fixes will be in.
> | >> >
> | >> >
> | >> >
> | >> > But does anything stand in the way of pressing the button on the
> | HP
> | >> > build, so that you have a HP 7.10.2 ready to go? You can always
> | >> > press the button again if/when further fixes go in.
> | >> >
> | >> >
> | >> >
> | >> > Simon
> | >> >
> | >> >
> | >> >
> | >> > *From:*ghc-devs [mailto:ghc-devs-bounces at haskell.org] *On Behalf
> | Of
> | >> > *Mark Lentczner
> | >> > *Sent:* 02 May 2015 01:57
> | >> > *To:* ghc-devs at haskell.org
> | >> > *Subject:* release timing
> | >> >
> | >> >
> | >> >
> | >> > I'm wondering if ghc 7.10.2 has a rough time table yet?
> | >> >
> | >> >
> | >> >
> | >> > This time 'round I'd like to do a HP 7.10.2 concurrent with GHC
> | 7.10.2!
> | >> > Crazy, I know, and who knows if we can pull it off....
> | >> >
> | >> >
> | >> >
> | >> > But I imagine it is about 3 or 4 week cycle for HP at this point
> | >> > (still)... so a few weeks' "heads up" would be good. Thoughts?
> | >> >
> | >> >
> | >> >
> | >> > - Mark
> | >> >
> | >> >
> | >> >
> | >> > _______________________________________________
> | >> > ghc-devs mailing list
> | >> > ghc-devs at haskell.org
> | >> > http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs
> | >> >
> | >>
> | >>
> | >>
> | >> _______________________________________________
> | >> ghc-devs mailing list
> | >> ghc-devs at haskell.org
> | >> http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs
> | >>
> | >
> | >
> | > _______________________________________________
> | > ghc-devs mailing list
> | > ghc-devs at haskell.org
> | > http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs
> | >
> | _______________________________________________
> | ghc-devs mailing list
> | ghc-devs at haskell.org
> | http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.haskell.org/pipermail/ghc-devs/attachments/20150505/c365b118/attachment.html>
More information about the ghc-devs
mailing list