I'm getting three or four validate failures on 64-bit Linux (openSuSE 12.3):
Unexpected failures:
perf/compiler T3064 [stat not good enough] (normal)
perf/compiler T6048 [stat not good enough] (optasm)
simplCore/should_compile simpl015 [exit code non-0] (optasm)
That would fix it in HEAD, but not 7.8, of course.
Cheers,
Simon
On 18/04/14 23:33, Simon Peyton Jones wrote:
Now we have forked the tree
- we can make Applicative a superclass of Monad
- and remove all the AMP warning stuff
That should resolve this particular issue, shouldn't it?
Austin, woul
I started preparing patches for #7858. My abs patch works well, my
signum patch not at all.
The gist of the bug is that abs and signum return erroneous values for
for the input -0.0. abs (-0.0 :: Float) should return 0.0, not -0.0. My
patch fixes this. For signum the situation is reverse. My patch
Make the `Float` and `Double` implementations of `signum` handle -0.0
correctly per IEEE-754.
---
libraries/base/GHC/Float.lhs | 16 ++--
libraries/base/changelog.md | 2 +-
2 files changed, 11 insertions(+), 7 deletions(-)
diff --git a/libraries/base/GHC/Float.lhs b/libraries/base/
Make the `Float` and `Double` implementations of `abs` handle -0.0
correctly per IEEE-754.
---
libraries/base/GHC/Float.lhs | 12
libraries/base/changelog.md | 2 ++
2 files changed, 10 insertions(+), 4 deletions(-)
diff --git a/libraries/base/GHC/Float.lhs b/libraries/base/GHC/Flo
-1 from me.
Your first example even provides a counter-example.
typedef enum {
> IMG_INIT_JPG = 0x0001,
> IMG_INIT_PNG = 0x0002,
> IMG_INIT_TIF = 0x0004,
> IMG_INIT_WEBP = 0x0008
> } IMG_InitFlags;
Those are defined as powers of two becaus
Have you considered a type class and new types ? :-)
On Saturday, April 19, 2014, Merijn Verstraaten
wrote:
> After some thinking it'd be possible to implement this using a quasiquoter
> and pattern synonyms, however you'd lose the ability to get exhaustiveness
> checks for the pattern synonyms.
On 2014-04-13 at 09:58:50 +0200, Herbert Valerio Riedel wrote:
[...]
> - base
> - ghc-prim
> - integer-gmp
> - integer-simple
> - template-haskell
[...]
> If no objections are raised, I'm planning to implement this change
> next weekend (April 19th/20th).
As there were no objections, I went on
On Saturday 19 April 2014, 14:53:04, Herbert Valerio Riedel wrote:
>
> Please try 'git submodule update --init' in the GHC source-tree top
> folder (that should be taken care of by sync-all, but sync-all wasn't
> adapted yet)
Not sure if it's progress, running that before 'sync-all get' fails wit
On 2014-04-19 at 14:08:19 +0200, Daniel Fischer wrote:
[...]
>> Klone nach 'libraries/xhtml'...
>> Fertig.
>> Unterprojekt-Pfad: 'libraries/xhtml':
>> 'fb9e0bbb69e15873682a9f25d39652099a3ccac1' ausgecheckt fatal: Projektarchiv
>> '/home/dafis/GHC/./haddock.git' existiert nicht. Klonen von
>> '/ho
Like so often, ./sync-all pull failed today in my local clones, first with
"... not on a branch", after sync-all checkout master, with "fatal: Reference
... not a tree". (Aside, that nonsense didn't happen before the introduction
of submodules.)
The usual remedy is
rm -rf ./clone
git clone ./g
After some thinking it'd be possible to implement this using a quasiquoter and
pattern synonyms, however you'd lose the ability to get exhaustiveness checks
for the pattern synonyms. Would people be equally opposed to the idea of
defining a "closed" set of synonyms? That is a user defines a set
On 15/04/14 05:52, Jan Stolarek wrote:
I think Simon Marlow is even more conservative, and
does his validate builds in a separate tree so he catches missing files.
I thought everyone is doing that.
It's the recommended workflow, but I must admit it doesn't always work
out, and nowadays I ofte
13 matches
Mail list logo
