[Toybox] A couple of minor fixes

[Rob Landley]

On 06/13/2016 02:12 AM, Mike Moreton wrote:
>> have you reported this bug to your supplier?
>>
>> have you experimented to work out what they largest value they
>> _can_ cope with is? does your version work with
>> INT_MAX? (maybe your supplier backported the "fix" that's
>> introduced a new bug?)
> 
> With the greatest of respect (genuinely), I think this is slightly
> missing the point - there are many ways I can fix this problem for
> us, so it's no great problem for us.
> 
> I was just reporting another configuration where it causes issues,
> because the more configurations there are where it causes issues,
> the greater the proportion of people trying out Toybox for the
> first time will reject it "because even grep doesn't work".
> 
> I'm sure we all agree that they'd be wrong to do that, but people
> trying a new package out for the first time tend to have a low
> tolerance level for issues.  I'm not saying our configuration is
> the straw that breaks the camel's back (I'll leave that decision
> to others), just that it's a straw.

It works on uClibc, it works on musl, it works on bionic, and it works
on older glibc (the version in ubuntu 12.04).

There was a bug introduced into newer glibc (ubuntu 14.04), which we
introduced a workaround for. Even new glibc made the breakage worse, and
our workaround for the original glibc breakage no longer works.

What I _can_ do is add a build-time check for this bug. It won't
work when cross compiling but I can catch it in native builds
and stop the build explaining how and why glibc broke.

That is a change I'm willing to make to toybox.

Rob