[OE-core] [PATCH 5/5] glibc: Dont offer to be parsed for non-glibc TCLIBC selection

[Khem Raj]

On Mon, Feb 23, 2015 at 7:00 AM, Christopher Larson <clarson at kergoth.com> wrote:
>
> On Thu, Feb 19, 2015 at 9:24 PM, Khem Raj <raj.khem at gmail.com> wrote:
>>
>> +    notglibc = (re.match('.*uclibc
>> , d.getVar('TARGET_OS', True)) != None) and (re.match('.*musl,
>> d.getVar('TARGET_OS', True)) != None)
>> +    if notglibc:
>
>
> Pretty sure this is a condition that will never be satisfied. This will only
> succeed if TARGET_OS is both musl and uclibc.

right ! I should have paid more attention. But since bitbake stopped
complaining with my test case
i relied on it not warning me. but I realise it should have just
reused the parser cache

Could we skip on TCLIBC !=
> glibc, to be both future proof and fix the logic?

Actually that was the patch I have done prior to this ( see
meta-linaro) but thats too restrictive since
it also meant that it started to ignore it for nativesdk-libc and we
use glibc as a provider to nativesdk version
of libc even in cases when target libc is musl or uclibc. So TARGET_OS
is really the correct variable here.

> --
> Christopher Larson
> clarson at kergoth dot com
> Founder - BitBake, OpenEmbedded, OpenZaurus
> Maintainer - Tslib
> Senior Software Engineer, Mentor Graphics