Re: Use gen-libm-test.py to generate ulps table for manual

[Joseph Myers <joseph at codesourcery dot com>]

On Thu, 16 Aug 2018, Michael Matz wrote:

> a reason to choose one or the other.  Rather all other reasons to prefer 
> one over the other should trump such personal preferences.  Such 
> considerations include: preexisting use of language $X, wideness of 
> availability for language $X interpreters/compilers and so on)

I think popularity of a language is a perfectly relevant consideration 
(not the only one, of course).  It's not simply among glibc developers 
that python is preferred to perl; various language rankings show python 
has much higher popularity than perl nowadays, which serves as external 
evidence that new contributors as well are more likely to be comfortable 
with build infrastructure in python than in perl.

> Another point I made over at gcc@ is that for base tools like gcc and 
> glibc the list of dependencies should be as minimal as possible, even at 

I think the considerations for glibc are separate from those for GCC 
(given, for example, that glibc is almost certainly built, even when being 
cross-compiled, on a GNU system for which a range of build tools are 
likely already available, including python, whereas GCC allows for being 
built natively on a non-GNU system that may not have such tools, as a 
starting point for then building such other tools which might not build 
with random vendor compilers on proprietary Unices).

Building glibc is also much more specialized than building GCC (if you're 
building glibc you're probably either a glibc developer or a system 
integrator using it as part of building a whole system, rather than just 
building a newer version locally for your own use, and using a different 
version of glibc than that coming with the system is tricky, whereas 
building your own GCC more recent than any that comes with the OS is an 
expected and supported normal use of GCC).

-- 
Joseph S. Myers
joseph@codesourcery.com