Hi Alex,

You're deleting hacks that were added for ancient version of binutils (added in 14a5aadb84c34dbe2bee129ed80fdfa1fb12e3e0 in 2005 and b8a64b8ec9cd3d8f6e3f23e44312c4903eccac45 in 2007).

I think that if you submit your patch without a test, there's a good chance it will get accepted.

Thomas

On Sat, Jul 9, 2016 at 3:22 PM, Edward Z. Yang <ezyang@mit.edu> wrote:

I am not sure if this will work, but how about dumping the assembly and
looking for sign extension? C-- might be easier!

Excerpts from Alex Dzyoba's message of 2016-07-09 08:25:39 -0400:

> Hi, all!
>
> I was working on #11758, which is about dropping binutils<2.17 hack, and while
> it was relatively easy to remove the hack itself, I'm not sure how to add a
> test case for it.
>
> As I understand, after removing the aforementioned hack, native codegen now
> shouldn't generate sign extension. So my question is how to test it? Should it
> be Cmm file that will be tested with `compile_cmp_asm` like memcpy in
> "codeGen/should_gen_asm/memcpy.cmm"? Or should I stick to
> the Haskell test?
>
> Thanks,
> Alex Dzyoba
_______________________________________________
ghc-devs mailing list
ghc-devs@haskell.org
http://mail.haskell.org/cgi-bin/mailman/listinfo/ghc-devs