Re: [PING][PATCH]: Warn that mpfr-2.2.0 is buggy when configuring

[Mike Stump <mrs at apple dot com>]

On Dec 28, 2006, at 5:56 AM, Paolo Bonzini wrote:
> I disagree with Joe on this: I think that we should not accept MPFR  
> 2.2.1 because, by the time GCC 4.3 is out, most distros will have  
> upgraded to the new version.

Please lay out the benefits and costs as you see them to breaking the  
automated testers and breaking the builds of all the developers and  
testers of gcc that have not updated to the version of the day that  
you'd like to require, and then, try and convince us that there is  
enough benefit for the cost involved.

Please predict how often you'd like to require us to update those  
libraries.  Once a day, once a month, once a release?

Personally, I don't see enough benefit to requiring the update now.   
I'd rather that the requirement be bumped on the 4.3 release branch  
once we are closer to release.  By that time, all the additional bugs  
that people find can be fixed and gotten into a release, and then, we  
only have to grab and build the libraries once a major release.  If  
you want to rev quickly, once a major release should be able to meet  
the quality goals you want to set for release.  Do you agree?  People  
that want to contribute to testing mpfr and gmp can updates their  
libraries nightly, if they want, some will.  I just ask that you not  
force people to updating more than once a major release.