
DarcsURL: http://darcs.haskell.org/ghc-6.8/packages/stm
MIME-Version: 1.0
Content-Type: multipart/mixed; boundary="=_"
--=_
Content-Type: text/plain
Content-Transfer-Encoding: quoted-printable
Thu Sep 20 21:56:33 =93=8C=8B=9E (=95W=8F=80=8E=9E) 2007 shelarcy

On Thu, Sep 20, 2007 at 05:59:55AM -0700, shelarcy wrote:
[Add dropped STM invariants functions shelarcy
**20070920125633] { hunk ./Control/Monad/STM.hs 27 + always, + alwaysSucceeds, }
I'm a little lost - what were these functions dropped from? They aren't exported by Control.Monad.STM in GHC 6.6.1 either. I'm not overly familiar with the STM package, but this seems to be an interface change, so I think it ought to go through http://www.haskell.org/haskellwiki/Library_submissions Thanks Ian

Hi Ian,
On Fri, 21 Sep 2007 00:18:32 +0900, Ian Lynagh
[Add dropped STM invariants functions shelarcy
**20070920125633] { hunk ./Control/Monad/STM.hs 27 + always, + alwaysSucceeds, } I'm a little lost - what were these functions dropped from?
I think they dropped from GHC 6.8 branch and GHC 6.8.1 release.
They aren't exported by Control.Monad.STM in GHC 6.6.1 either.
They are GHC 6.8.x's new feature. So current GHC 6.6.1 and stm packages can't export them. http://www.haskell.org/pipermail/haskell/2006-October/018618.html I thought someone add them in GHC 6.7 developping phase. But anyone doesn't do it, and now we are in the GHC 6.8.1 Release Candidate phase. So, I send patch to add thses functions. I think this is important change for STM user. Because if these functions are dropped from GHC 6.8.1' release, they must also use GHC.Conc module instead of just using Control.Concurrent.STM (or Control.Monad.STM). And this causes same tragedy on GHC 6.6.x's Parallel Haskell again. Parallel Haskell dropped exporting pseq function when GHC 6.6 release time, so we must use bad interface until GHC 6.8.1 release. http://www.mail-archive.com/glasgow-haskell-users@haskell.org/msg10973.html
I'm not overly familiar with the STM package, but this seems to be an interface change, so I think it ought to go through http://www.haskell.org/haskellwiki/Library_submissions
Should I formalise this contribution? I thought STM package developper knows GHC 6.8.x's change, and just forgot adding functions to STM package. So I send just darcs patch. Best Regards. -- shelarcy <shelarcy hotmail.co.jp> http://page.freett.com/shelarcy/

I think shelarcy is right here. The point of release candidates is to identify bugs, including API bugs, in the release candidate. We freeze the API once we release.
In this case, the functionality is implemented, and its API should be in the main STM API.
thanks for catching this shelarcy!
Simon
| -----Original Message-----
| From: libraries-bounces@haskell.org [mailto:libraries-bounces@haskell.org] On Behalf Of shelarcy
| Sent: 20 September 2007 17:52
| To: Ian Lynagh
| Cc: libraries@haskell.org
| Subject: Re: darcs patch: Add dropped STM invariants functions
|
| Hi Ian,
|
| On Fri, 21 Sep 2007 00:18:32 +0900, Ian Lynagh

Hi shelarcy, On Fri, Sep 21, 2007 at 01:52:19AM +0900, shelarcy wrote:
On Fri, 21 Sep 2007 00:18:32 +0900, Ian Lynagh
wrote: [Add dropped STM invariants functions shelarcy
**20070920125633] { hunk ./Control/Monad/STM.hs 27 + always, + alwaysSucceeds, } http://www.haskell.org/pipermail/haskell/2006-October/018618.html
Ah, I see, some new functionality is not exported from the preferred module. OK, I've applied the patch - thanks for sending it! Thanks Ian
participants (3)
-
Ian Lynagh
-
shelarcy
-
Simon Peyton-Jones