Re: [PATCH v2 03/10] i386: Use generic fmod

[Adhemerval Zanella Netto <adhemerval.zanella@linaro.org>]

On 28/03/24 15:38, Joseph Myers wrote:
> On Thu, 28 Mar 2024, Adhemerval Zanella Netto wrote:
> 
>> We don't really have SVID compatibility tests for any other 
>> optimization/simplification, and although I don't really oppose on 
>> adding I also thinking that this is making this change even more 
>> complicated than it would require.
> 
> We have math/test-matherr* as generic tests of matherr working with 
> binaries that specifically use the compat symbol matherr and not with 
> newly linked binaries.  Those don't however attempt to test properties of 
> specific symbols for individual libm functions.

I meant that we don't have explicit SVID tests that check its semantic
(including return code, matherr status, stdout return code), like we do
for some compat symbols.  And I took that this what H.J is asking about.

> 
> A change like the present one is not meant to be user-visible for any 
> newly linked binary, as a legitimate newly linked binary couldn't link 
> against the matherr compat symbol anyway.  And we don't have any attempt 
> to ensure that old symbol versions of individual libm functions can be 
> used together with the compat symbol of matherr - just that the compat 
> symbol of matherr works with one libm function.
>