Backporting srcLoc to the GHC 7.10 branch

Michael Snoyman michael at snoyman.com
Tue Mar 24 07:00:02 UTC 2015


Thanks for the feedback everyone. If/when we decide to backport the patch,
we'll be sure to make the commit available for others interested in running
a custom build of GHC. And thank you for the offer of help Eric, it's much
appreciated :)

On Mon, Mar 23, 2015 at 8:14 PM Eric Seidel <eric at seidel.io> wrote:

> IIRC the patch doesn't directly depend on -fwarn-redundant-constraints,
> but I think I ran into some merge conflicts that had to be resolved. I
> agree with Austin that backporting it should be doable, and would be
> happy to help if there's interest.
>
> Also, I still mean to submit a follow-on patch that uses the new srcLoc
> infrastructure to add locations to explicitly-failing functions (i.e.
> error, undefined, and maybe assert). I don't know if this will be
> palatable to GHC-HQ as it changes base, but I think it's at least a
> discussion worth having. Unfortunately I got sidetracked by school stuff
> and haven't had a chance to throw the patch together yet..
>
> On Mon, Mar 23, 2015, at 10:50, Austin Seipp wrote:
> > Hi Michael,
> >
> > I actually tried to adopt this patch into 7.10, because I thought it
> > was needed for another dependent patch we wanted. Unfortunately, the
> > backtrace-via-implicit-params patch seems to depend on some prior work
> > by Simon PJ in the typechecker (-fwarn-redundant-constraints, a rather
> > large patch), which we *did not* want in 7.10 (the *textual* diff
> > applied fine, but there were some API changes the backtrace patch
> > needed, so it failed to build). So, in the end, it was easier to
> > surgically remove this patch from the one that depended on it, and it
> > had a much lower 'surface area' of changes, than adopting both. Hope
> > that makes sense.
> >
> > Also, as Simon said, we don't normally do big changes like this in
> > point releases, so I think this is unlikely to happen.
> >
> > So the short story is "afraid not". But a backport should be possible,
> > if you're willing to get your hands a bit dirty (any conflict will
> > likely be fairly easy to fix, but it will involve some textual
> > munging).
> >
> > On Mon, Mar 23, 2015 at 6:49 AM, Michael Snoyman <michael at snoyman.com>
> > wrote:
> > > It looks like the srcLoc change[1] is something that some of our (FP
> > > Complete's) customers would be quite interested in getting access to
> sooner
> > > rather than later. Would there be any possibility of getting that patch
> > > merged into the 7.10 branch of GHC? I'm not sure if I'd try my luck at
> > > actually including it in 7.10.1, but would it be on the table for
> 7.10.2?
> > >
> > > We do of course have the option of backporting it ourselves and
> including it
> > > in a custom GHC we provide customers, but we generally try to stay as
> close
> > > to upstream as possible.
> > >
> > > [1] https://ghc.haskell.org/trac/ghc/ticket/9049
> > >
> > > _______________________________________________
> > > ghc-devs mailing list
> > > ghc-devs at haskell.org
> > > http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs
> > >
> >
> >
> >
> > --
> > Regards,
> >
> > Austin Seipp, Haskell Consultant
> > Well-Typed LLP, http://www.well-typed.com/
> > _______________________________________________
> > 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/20150324/40b74798/attachment.html>


More information about the ghc-devs mailing list