remaining gnulib/glibc incompatibilities

remaining gnulib/glibc incompatibilities

[John Spencer]

i tested with m4, using my configure.cache workaround ( 
https://github.com/rofl0r/sabotage/blob/master/KEEP/config.cache#L77 ) 
for bogus tests:

the following configure tests failed, apart from getopt, which is 
expected; causing a gnulib replacement to get built:

http://sprunge.us/FcJT putenv expects GNU behaviour,

http://sprunge.us/DdhN regex expects a ton of GNU additions, causing 
about 120KB in objects to get built,

http://sprunge.us/EfdS and strtod expects "nan()" to be accepted.

Re: remaining gnulib/glibc incompatibilities

[Szabolcs Nagy]

* John Spencer <maillist-musl@barfooze.de> [2012-10-16 19:51:42 +0200]:
> http://sprunge.us/EfdS and strtod expects "nan()" to be accepted.

this one can be easily fixed

(it seems gratuitous complexity in the standard, but that's how it is)

i don't know the internals of floatscan though
i guess the 'pok' argument tells if putback is ok
and if it's false then at most one putback is allowed

if arbitrary long putback is possible when pok==1
then i can create a patch (that always returns the
same nan for all correct nan(...) formats)

otherwise we can add support for "nan" and "nan()" only

Re: remaining gnulib/glibc incompatibilities

[Rich Felker]

On Tue, Oct 16, 2012 at 07:51:42PM +0200, John Spencer wrote:
> i tested with m4, using my configure.cache workaround (
> https://github.com/rofl0r/sabotage/blob/master/KEEP/config.cache#L77
> ) for bogus tests:
> 
> the following configure tests failed, apart from getopt, which is
> expected; causing a gnulib replacement to get built:
> 
> http://sprunge.us/FcJT putenv expects GNU behaviour,

Fixed in git.

> http://sprunge.us/DdhN regex expects a ton of GNU additions, causing
> about 120KB in objects to get built,

Not really fixable.

> http://sprunge.us/EfdS and strtod expects "nan()" to be accepted.

Fixed in git.

Rich