I upgraded to Ubuntu 10.10. So far, it seems to work, but the std.conv unit tests for Phobos1 are failing. Seems that strtof() doesn't work the same. Argh.

On 1/21/2011 8:38 PM, Walter Bright wrote:
> I upgraded to Ubuntu 10.10. So far, it seems to work, but the std.conv unit
> tests for Phobos1 are failing. Seems that strtof() doesn't work the same. Argh.
> _______________________________________________
> phobos mailing list
> phobos at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/phobos

My linux 64 bit box is running 10.10.. not that helps a lot as I do primarily 64 bit development.  I rarely run 32 bit builds though.  I'll do a 32 bit test pass over d1 and d2 on it right now.

On 1/21/2011 10:21 PM, Brad Roberts wrote:
> On 1/21/2011 8:38 PM, Walter Bright wrote:
>> I upgraded to Ubuntu 10.10. So far, it seems to work, but the std.conv unit tests for Phobos1 are failing. Seems that strtof() doesn't work the same. Argh.
> 
> My linux 64 bit box is running 10.10.. not that helps a lot as I do primarily 64 bit development.  I rarely run 32 bit builds though.  I'll do a 32 bit test pass over d1 and d2 on it right now.

d1 32 bit's phobos tests does fail with:
  Error: conversion 1.17549e-38 not representable as a float

d2 32 bit's phobos tests are clean.

Brad Roberts wrote:
> On 1/21/2011 10:21 PM, Brad Roberts wrote:
> 
>> On 1/21/2011 8:38 PM, Walter Bright wrote:
>> 
>>> I upgraded to Ubuntu 10.10. So far, it seems to work, but the std.conv unit
>>> tests for Phobos1 are failing. Seems that strtof() doesn't work the same. Argh.
>>> 
>> My linux 64 bit box is running 10.10.. not that helps a lot as I do primarily 64
>> bit development.  I rarely run 32 bit builds though.  I'll do a 32 bit test pass
>> over d1 and d2 on it right now.
>> 
>
> d1 32 bit's phobos tests does fail with:
>   Error: conversion 1.17549e-38 not representable as a float
>
> d2 32 bit's phobos tests are clean.
>
> 

Yup, just what I'm getting.

Walter Bright wrote:
> 
> Yup, just what I'm getting.
>

arrayops.d is failing now, too. It seems that the new printf() trashes the XMM0 register, which breaks the code gen. I'll look into fixing it tomorrow.