Re: _Float128 patches was: Defining macros for function symbol versions

[Joseph Myers <joseph at codesourcery dot com>]

On Fri, 9 Dec 2016, Tulio Magno Quites Machado Filho wrote:

> I've just pushed those patches to tuliom/float128.
> They aren't ready and I don't expect people to review them before they're
> submitted to this mailing list.
> 
> However, if you're planning to work on _Float128, taking a look at this code
> would help to avoid duplicated efforts.

Thanks.  I see nothing there that would conflict with TS 18661-1 work 
beyond trivial mechanical updates for new functions.  I see nothing that 
would conflict with preparations for TS 18661-3 function aliases, or with 
such aliases (which I don't expect to start on for 2.25 anyway) for types 
other than _Float128.

It's reasonable for a float128 patch to be posted based on a slightly 
outdated version of mainline (a day or two old, so missing functions added 
in the past day or two but properly covering older functions), as long as 
it's clear what version of mainline the patch is based on and the known 
changes since then are only relevant in mechanical ways rather than 
needing more significant changes to the posted patch (or if more 
significant changes are needed, it's explicitly posted for review of other 
aspects of the patch).

If at some point we have float128 patches ready for mainline except for 
being based on sources that are a day or two old, at that point I'd 
consider it reasonable to approve them subject to mechanical updates for 
new functions and hold off on additions of other libm functions for a day 
or two for the final rebasing and testing and commit to take place.  
Right now, we are not at that point, as patches need more work than just 
mechanically handling very new functions in various places.

-- 
Joseph S. Myers
joseph@codesourcery.com