Feed Advanced Search

Nov 23 2014

carter updated the diff for D350: Changing prefetch primops to have a `seq`-like interface.
  • the prefetchValue operations should be lazy in their value argument
Nov 23 2014, 11:17 AM
carter added a comment to D350: Changing prefetch primops to have a `seq`-like interface.

i'm concerned that the prefetchValue operations dont have the right strictness information on them. The value argument shouldn't be evaluated.

Nov 23 2014, 10:42 AM
carter updated the diff for D350: Changing prefetch primops to have a `seq`-like interface.
  • Revert "somehow how i modified the docs made the primops file parser fail. twiddling that"
  • Revert "clarify documention to make it clear that inlinePerformIO is safe to use here"
  • clarify documentation about prefetch
Nov 23 2014, 2:07 AM
carter updated the diff for D350: Changing prefetch primops to have a `seq`-like interface.
  • somehow how i modified the docs made the primops file parser fail. twiddling that
Nov 23 2014, 1:57 AM
carter updated the diff for D350: Changing prefetch primops to have a `seq`-like interface.
  • clarify documentation to make it clear that inlinePerformIO is safe to use here
Nov 23 2014, 1:45 AM
carter added a comment to D350: Changing prefetch primops to have a `seq`-like interface.

agreed, should I change the recommendation to unsafeDupable or inlinePerformIO?

Nov 23 2014, 1:32 AM
carter added a comment to D519: preliminary Complex and Ratio Storable instances for Base.

and just to be clear, it has a test!

Nov 23 2014, 1:18 AM
carter updated the diff for D350: Changing prefetch primops to have a `seq`-like interface.
  • fixing up some validate -werror noise
Nov 23 2014, 1:07 AM
carter added a comment to D519: preliminary Complex and Ratio Storable instances for Base.

wooot, it validates

Nov 23 2014, 1:06 AM
carter updated the diff for D350: Changing prefetch primops to have a `seq`-like interface.
  • remove some deadcode from the test
  • woops, mixed up arg types in primop helpers
Nov 23 2014, 12:41 AM
carter updated the diff for D519: preliminary Complex and Ratio Storable instances for Base.
  • use the realPart function to avoid needing the scoped type variables extension
Nov 23 2014, 12:39 AM
carter added a reviewer for D350: Changing prefetch primops to have a `seq`-like interface: ekmett.

Ed, Simon, feedback would be appreciated

Nov 23 2014, 12:34 AM
carter updated the diff for D350: Changing prefetch primops to have a `seq`-like interface.
  • updating prefetch to have a better api
Nov 23 2014, 12:32 AM
carter updated the diff for D519: preliminary Complex and Ratio Storable instances for Base.
  • rewrote test to use operations exposed in base lib
  • making the instances correctly lazy in the proxy args
Nov 23 2014, 12:24 AM

Nov 22 2014

carter added a comment to D519: preliminary Complex and Ratio Storable instances for Base.

having trouble getting validate to work locally on my mac, but trying to track down whats wrong with the test suite

Nov 22 2014, 11:50 PM
carter added a comment to D519: preliminary Complex and Ratio Storable instances for Base.

for other reasons validate seems to be broken on os x, so i'm having a bit of trouble tracking down this bug

Nov 22 2014, 11:32 PM
carter updated the diff for D519: preliminary Complex and Ratio Storable instances for Base.
  • woops, typo in test suite
Nov 22 2014, 10:50 PM
carter updated the diff for D519: preliminary Complex and Ratio Storable instances for Base.
  • futzing with stdout
Nov 22 2014, 10:47 PM
carter updated the diff for D519: preliminary Complex and Ratio Storable instances for Base.
  • using Foreign rather than Foreign.Safe
Nov 22 2014, 9:19 PM
carter updated the diff for D350: Changing prefetch primops to have a `seq`-like interface.

twiddle

Nov 22 2014, 9:16 PM
carter updated the diff for D350: Changing prefetch primops to have a `seq`-like interface.
  • updated the documention about how to use pure prefetch. also fixed a type error in the in the test
  • white space wibble
  • for the example in the documentation, replacing
  • cargo cult changing the formatting to see if that makes a difference
  • revert removing the example, also try wrapping the result in (# a #)
  • adjusting test suite for new type signature for pure prefetchs
Nov 22 2014, 9:13 PM
carter updated the diff for D519: preliminary Complex and Ratio Storable instances for Base.
  • derp, forgot to import storable class
Nov 22 2014, 8:25 PM
carter added a comment to D519: preliminary Complex and Ratio Storable instances for Base.

Fixed up the Ratio instance with the % and Integral a, as suggested,

Nov 22 2014, 4:09 PM
carter updated the diff for D519: preliminary Complex and Ratio Storable instances for Base.
  • shifting to using Foreign.Safe to make it clearer that Data.Complex stays safe
Nov 22 2014, 4:08 PM
carter updated the diff for D519: preliminary Complex and Ratio Storable instances for Base.
  • make sure invariants of ratio are maintained on peaks
Nov 22 2014, 4:03 PM
carter updated the diff for D519: preliminary Complex and Ratio Storable instances for Base.
  • adding basic test case for ratio and complex storable instances
Nov 22 2014, 3:55 PM
carter updated the diff for D519: preliminary Complex and Ratio Storable instances for Base.
  • add changelog notes and tweak Ratio instance typesetting
Nov 22 2014, 2:03 PM
carter updated the diff for D519: preliminary Complex and Ratio Storable instances for Base.
  • drop using the realPart operations to get the fields
Nov 22 2014, 11:47 AM
carter updated the diff for D519: preliminary Complex and Ratio Storable instances for Base.
  • fixup parsing error in definitions
Nov 22 2014, 10:05 AM

Nov 21 2014

carter retitled D519: preliminary Complex and Ratio Storable instances for Base from to preliminary Complex and Ratio Storable instances for Base.
Nov 21 2014, 7:51 PM

Nov 18 2014

carter added a comment to D496: trac #9744, make program name and product version configurable through DynFlags/Settings.

naively it looks fine to me.

Nov 18 2014, 4:09 PM
carter added a comment to D486: Restore exact old semantics of `decodeFloat`.

hrm, how come on my 7.8.3 ghci i'm seeing the the following

Nov 18 2014, 1:24 PM

Nov 7 2014

carter added a comment to D432: Improve Applicative definitions.

this look great. nice work

Nov 7 2014, 12:57 AM

Nov 5 2014

carter added a comment to D165: driver: Add support for intel/amd -march/mcpu flags.

whats needed to help get this into a mergeable state? I'm willing to do some leg work if any is needed!

Nov 5 2014, 9:08 AM

Nov 4 2014

carter added inline comments to D428: Implement folds for Array directly.
Nov 4 2014, 9:48 PM · GHC

Oct 30 2014

carter added a comment to D155: llvmGen: Compatibility with LLVM 3.5.

Huh, that might explain some hard to reproduce bugs i've hit when building split objects code on OS X.

Oct 30 2014, 12:14 AM

Oct 20 2014

carter added a comment to D252: Support C structure as return value of ccall in cmm.

@tibbe @simonmar, should there be a stack check before allocating there?

Oct 20 2014, 1:30 PM · GHC
carter added a comment to D252: Support C structure as return value of ccall in cmm.

I"m noticing that

Oct 20 2014, 1:28 PM · GHC
carter added a comment to D353: Make calling conventions in template haskell Syntax.hs consistent with those in ghc ForeignCall.hs this impliments #9703 from ghc trac.

Every build of GHC right now is failing that perf test. I think austin or
hvr are going to tweek it later this week. But if not maybe that's a good
second patch !

Oct 20 2014, 12:35 AM

Oct 18 2014

carter added a comment to D350: Changing prefetch primops to have a `seq`-like interface.

looks like the only failure in the build is perf/compiler T3064 [stat not good enough] (normal)
Which is not my test! :)

Oct 18 2014, 11:51 PM
carter updated the diff for D350: Changing prefetch primops to have a `seq`-like interface.
  • adjusting test suite for new type signature for pure prefetchs
Oct 18 2014, 10:43 PM
carter updated the diff for D350: Changing prefetch primops to have a `seq`-like interface.
  • revert removing the example, also try wrapping the result in (# a #)
Oct 18 2014, 9:51 PM
carter updated the diff for D350: Changing prefetch primops to have a `seq`-like interface.
  • cargo cult changing the formatting to see if that makes a difference
Oct 18 2014, 9:08 PM
carter updated the diff for D350: Changing prefetch primops to have a `seq`-like interface.
  • changing the example in the docs because it was somehow triggering a GHC panic in a way I dont understand (possibly via triggering a parser error)
Oct 18 2014, 8:26 PM
carter added a comment to D350: Changing prefetch primops to have a `seq`-like interface.

i'm trying to understand the compiler panic i'm getting, but honestly quite stumped

Oct 18 2014, 8:07 PM
carter updated the diff for D350: Changing prefetch primops to have a `seq`-like interface.
  • white space wibble
Oct 18 2014, 6:03 PM
carter updated the diff for D350: Changing prefetch primops to have a `seq`-like interface.
  • updated the documention about how to use pure prefetch. also fixed a type error in the in the test
Oct 18 2014, 6:00 PM
carter updated the diff for D350: Changing prefetch primops to have a `seq`-like interface.
  • updated the documention about how to use pure prefetch. also fixed a type error in the test
Oct 18 2014, 4:31 PM
carter added a comment to D350: Changing prefetch primops to have a `seq`-like interface.

I need to update the docs to reflect the new seq style.

Oct 18 2014, 4:15 PM
carter retitled D350: Changing prefetch primops to have a `seq`-like interface from to Changing the pure prefetch primops to have seqlike behavior so that prefetch can be used effectively in pure codes! This is based upon design discused on trac ticket #9353.
Oct 18 2014, 3:44 PM

Oct 8 2014

carter added a comment to D252: Support C structure as return value of ccall in cmm.

I dont think we should assume 32bit mac is dead, or at least there needs to be a real formal dialogue about that.

Oct 8 2014, 10:39 AM · GHC

Sep 25 2014

carter added a comment to D244: Fixes cyclic import on OS X(#9635).

(tested on my OSX 10.9 mac)

Sep 25 2014, 11:10 PM
carter added a comment to D244: Fixes cyclic import on OS X(#9635).

LGTM. Just tested it out myself and the build unbreaks with that change

Sep 25 2014, 11:10 PM

Aug 24 2014

carter added a comment to D172: Remove Cabal dependency from GHC.

Question thats related: is Duncan's WIP new binary major version out yet? (or is that unrelated to the use of Binary in ghc?)

Aug 24 2014, 6:06 PM

Aug 20 2014

carter added a comment to D119: StaticValues language extension.

on a bike shedding note, before this gets merged in, could we please change all the notes/comments/wiki pages to NOT calling it static values? I'm (slowly) working on something that would be about using compile time fixed (static) values, and as has been noted here, this is more like "multiple compilations stable module name/handles"

Aug 20 2014, 12:12 PM

Aug 16 2014

carter added a comment to D142: use CPPHS when it is available unless specific CPP is set using --with-hs-cpp configure option.

I'm probably being a bit overconservative. Let me mull this a teeny bitmore + test if i can build ghc HEAD with cpphs on my mac in a clean room setting using cpphs, I remember hitting some issues with that before, because of weird spots where GCC is hard coded in.

Aug 16 2014, 3:55 PM
carter added a comment to D142: use CPPHS when it is available unless specific CPP is set using --with-hs-cpp configure option.

I'm a bit confused, need/should we even bake cpphs in? (if nothing else, the gymnastics of properly understanding the subtlties of these software licenses is an overhead that is both tricky and in general requires thoughtful collaboration with a literate lawyer)

Aug 16 2014, 12:17 PM

Aug 12 2014

carter added a comment to D145: Make Prelude.abs handle -0.0 correctly (#7858).

look good to me! This also brings our definition of abs in alignment with IEEE 754 2008 section 5.55
"abs(x) copies a floating-point operand x to a destination in the same format, setting the sign bit to 0 (positive)."

Aug 12 2014, 11:40 AM

Jul 3 2014

carter added a comment to D38: NB. this is an unfinished patch set that I'm putting up for people to see and/or help with. I believe it works apart from the byte-code backend. There are several commits squashed together here. WIP: Pack non-pointer fields of constructors.

could this be generalized to work with current or future iteration of the SIMD data types? (don't know if that quite makes sense, but thats my immediate question, though i'm still working through reading this patch)

Jul 3 2014, 1:55 PM

Jun 29 2014

carter added a comment to D26: Control CPP through settings file (#8683).

@ simomar, the fp_foo_bar bit was cruft from an earlier revision of this that suck into the diff i used to generate the cleaned up patch set, if you look at the most recent revision, there should be no mention of it in the diff

Jun 29 2014, 7:54 PM
carter added a comment to D26: Control CPP through settings file (#8683).

mark -- could you try doing the same process but without doing the clang wrapper?
on systems with clang that predate Xcode5, clang CPP WONT WORK. so eg on 10.7, you'll actually want to either use the provided GCC for both C compilation and CPP, or separately set the CPP.

Jun 29 2014, 1:32 PM

Jun 27 2014

carter updated the diff for D26: Control CPP through settings file (#8683).
  • Control CPP through settings file (Trac #8683)
  • remove dead FP_CPP_FLAGS_WITH_VAL_DEFAULTED
  • make the variable names more consistent for the Haskell CPP settings detection in the configure scripts
Jun 27 2014, 5:39 PM
carter added a comment to D26: Control CPP through settings file (#8683).

i created a RAWCPP ticket a while ago https://ghc.haskell.org/trac/ghc/ticket/9094

Jun 27 2014, 5:30 PM

Jun 26 2014

carter added a comment to D26: Control CPP through settings file (#8683).

one thing I should probably add on top of this patch if its not merged in for 7.8.3 is eliminating all appearances of RAWCPP_FLAGS in the entire ghc code base

Jun 26 2014, 4:02 PM

Jun 25 2014

carter added a comment to D26: Control CPP through settings file (#8683).

Ran an equivalent version of this patch through validate on the 7.8 branch on both OS X and linux yesterday
Unexpected results from:
TEST="T3064 T5837 T6048"

Jun 25 2014, 2:33 PM
carter retitled D26: Control CPP through settings file (#8683) from to Control CPP through settings file (#8683).
Jun 25 2014, 2:23 PM

Jun 9 2014

carter added a comment to Image Macro "crazy".

Macro crazy: testing

Jun 9 2014, 2:21 PM
carter updated subscribers of D15: Add more primops for atomic ops on byte arrays.
Jun 9 2014, 12:54 PM

Jun 5 2014

carter added a comment to D4: Add support for faster copies via Intel Extended REP-MOVSB.

I like this! (the march tooling and this workflow)

Jun 5 2014, 11:17 PM