'LIBC' variable clashes with perl variable

'LIBC' variable clashes with perl variable

[Koen Kooi]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi,

Perl (among others) uses the 'LIBC' variable to point its linker to
libc.so.6, which fails due to OE redefing LIBC.

Any objections to renaming LIBC to TARGET_LIBC in OE?

regards,

Koen
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.5 (Darwin)

iD8DBQFMOcrcMkyGM64RGpERAgPdAJ9DGZStg8DuhSa2MrYNuLsJVWRjwQCfZEPN
YIxL9q+S4caNMNmlrYsb3EE=
=xzeu
-----END PGP SIGNATURE-----

Re: 'LIBC' variable clashes with perl variable

[Tom Rini]

Koen Kooi wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Hi,
> 
> Perl (among others) uses the 'LIBC' variable to point its linker to
> libc.so.6, which fails due to OE redefing LIBC.
> 
> Any objections to renaming LIBC to TARGET_LIBC in OE?

Sounds sane to me.

-- 
Tom Rini
Mentor Graphics Corporation

Re: 'LIBC' variable clashes with perl variable

[Tom Rini]

Tom Rini wrote:
> Koen Kooi wrote:
>> -----BEGIN PGP SIGNED MESSAGE-----
>> Hash: SHA1
>>
>> Hi,
>>
>> Perl (among others) uses the 'LIBC' variable to point its linker to
>> libc.so.6, which fails due to OE redefing LIBC.
>>
>> Any objections to renaming LIBC to TARGET_LIBC in OE?
> 
> Sounds sane to me.

Actually, why don't we just unexport it, like MACHINE / DISTRO / 
TARGET_ARCH, and while we're at it, maybe TMPDIR too (since that's quite 
meaningful outside of OE).

-- 
Tom Rini
Mentor Graphics Corporation

Re: 'LIBC' variable clashes with perl variable

[Tom Rini]

Tom Rini wrote:
> Tom Rini wrote:
>> Koen Kooi wrote:
>>> -----BEGIN PGP SIGNED MESSAGE-----
>>> Hash: SHA1
>>>
>>> Hi,
>>>
>>> Perl (among others) uses the 'LIBC' variable to point its linker to
>>> libc.so.6, which fails due to OE redefing LIBC.
>>>
>>> Any objections to renaming LIBC to TARGET_LIBC in OE?
>>
>> Sounds sane to me.
> 
> Actually, why don't we just unexport it, like MACHINE / DISTRO / 
> TARGET_ARCH, and while we're at it, maybe TMPDIR too (since that's quite 
> meaningful outside of OE).

ping?

-- 
Tom Rini
Mentor Graphics Corporation

Re: 'LIBC' variable clashes with perl variable

[Koen Kooi]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 28-07-10 02:48, Tom Rini wrote:
> Tom Rini wrote:
>> Tom Rini wrote:
>>> Koen Kooi wrote:
>>>> -----BEGIN PGP SIGNED MESSAGE-----
>>>> Hash: SHA1
>>>>
>>>> Hi,
>>>>
>>>> Perl (among others) uses the 'LIBC' variable to point its linker to
>>>> libc.so.6, which fails due to OE redefing LIBC.
>>>>
>>>> Any objections to renaming LIBC to TARGET_LIBC in OE?
>>>
>>> Sounds sane to me.
>>
>> Actually, why don't we just unexport it, like MACHINE / DISTRO /
>> TARGET_ARCH, and while we're at it, maybe TMPDIR too (since that's
>> quite meaningful outside of OE).
> 
> ping?

Sounds like a good idea to me.

regards,

Koen
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.5 (Darwin)

iD8DBQFMT9erMkyGM64RGpERAhw6AKCVTO1RK3XH3B2YOekgEdvHYRRKBwCeNx+S
E0WYpLLZov9c/ZVmErVoZJ0=
=G0RS
-----END PGP SIGNATURE-----

Re: 'LIBC' variable clashes with perl variable

[Roman I Khimov]

В сообщении от Среда 28 июля 2010 04:48:19 автор Tom Rini написал:
> Tom Rini wrote:
> > Actually, why don't we just unexport it, like MACHINE / DISTRO /
> > TARGET_ARCH, and while we're at it, maybe TMPDIR too (since that's quite
> > meaningful outside of OE).
> 
> ping?

I don't see any problems with that, so why not?

Re: 'LIBC' variable clashes with perl variable

[Phil Blundell]

On Sun, 2010-07-11 at 15:45 +0200, Koen Kooi wrote:
> Perl (among others) uses the 'LIBC' variable to point its linker to
> libc.so.6, which fails due to OE redefing LIBC.
> 
> Any objections to renaming LIBC to TARGET_LIBC in OE?

How is perl getting to see that variable?  I don't think it ought to be
exported.

I don't have any problem with renaming it, though I do wonder whether it
would be better to just abolish it altogether in favour of inspecting
${HOST_OS}.

p.

Re: 'LIBC' variable clashes with perl variable

[Koen Kooi]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 11-07-10 19:01, Phil Blundell wrote:
> On Sun, 2010-07-11 at 15:45 +0200, Koen Kooi wrote:
>> Perl (among others) uses the 'LIBC' variable to point its linker to
>> libc.so.6, which fails due to OE redefing LIBC.
>>
>> Any objections to renaming LIBC to TARGET_LIBC in OE?
> 
> How is perl getting to see that variable?  I don't think it ought to be
> exported.
> 
> I don't have any problem with renaming it, though I do wonder whether it
> would be better to just abolish it altogether in favour of inspecting
> ${HOST_OS}.

LIBC is currently used to set it for bitbake, e.g.:

MACHINE=beagleboard LIBC=eglibc bitbake foo
MACHINE=beagleboard LIBC=uclibc bitbake foo

Which also means we can't use any '_', so it would be TARGETLIBC or HOSTOS.
I'm not sure what the impact would be when we would unexport it, though.

regards,

Koen
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.5 (Darwin)

iD8DBQFMOgBTMkyGM64RGpERAqO9AJ4lSJbHDviQjgkIK5XkTC6/9voHIQCcDSrq
CK72iQi4N6rTh7x9hXt3jpM=
=XkZF
-----END PGP SIGNATURE-----

Re: 'LIBC' variable clashes with perl variable

[Khem Raj]

On Sun, Jul 11, 2010 at 10:01 AM, Phil Blundell <philb@gnu.org> wrote:
> On Sun, 2010-07-11 at 15:45 +0200, Koen Kooi wrote:
>> Perl (among others) uses the 'LIBC' variable to point its linker to
>> libc.so.6, which fails due to OE redefing LIBC.
>>
>> Any objections to renaming LIBC to TARGET_LIBC in OE?
>
> How is perl getting to see that variable?  I don't think it ought to be
> exported.
>
> I don't have any problem with renaming it, though I do wonder whether it
> would be better to just abolish it altogether in favour of inspecting
> ${HOST_OS}.

uclibc is ok but glibc and eglibc will have same HOST_OS so it will be
hard to differentiate which system C library to choose.


>
> p.
>
>
>
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
>

Re: 'LIBC' variable clashes with perl variable

[Phil Blundell]

On Sun, 2010-07-11 at 15:17 -0700, Khem Raj wrote:
> uclibc is ok but glibc and eglibc will have same HOST_OS so it will be
> hard to differentiate which system C library to choose.

Are there any DISTROs which actually want to support both eglibc and
glibc in parallel?

p.

Re: 'LIBC' variable clashes with perl variable

[Khem Raj]

On Sun, Jul 11, 2010 at 3:39 PM, Phil Blundell <philb@gnu.org> wrote:
> On Sun, 2010-07-11 at 15:17 -0700, Khem Raj wrote:
>> uclibc is ok but glibc and eglibc will have same HOST_OS so it will be
>> hard to differentiate which system C library to choose.
>
> Are there any DISTROs which actually want to support both eglibc and
> glibc in parallel?

angstrom supports all 3 of them IIRC with defaults being glibc

>
> p.
>
>
>
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
>

Re: 'LIBC' variable clashes with perl variable

[Tom Rini]

Koen Kooi wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Hi,
> 
> Perl (among others) uses the 'LIBC' variable to point its linker to
> libc.so.6, which fails due to OE redefing LIBC.
> 
> Any objections to renaming LIBC to TARGET_LIBC in OE?

bitbake.conf: Unexport LIBC from the environment

As spoted by Koen at least perl makes use of LIBC coming from the 
environment to be meaningful, so don't export it as it doesn't match our 
meaning.

Signed-off-by: Tom Rini <tom_rini@mentor.com>

diff --git a/conf/bitbake.conf b/conf/bitbake.conf
index d523519..ddb6a79 100644
--- a/conf/bitbake.conf
+++ b/conf/bitbake.conf
@@ -799,3 +799,6 @@ TARGET_ARCH[unexport] = "1"
  # (breaks sysvinit at least)
  DISTRO[unexport] = "1"

+# Make sure that LIBC isn't exported
+# (breaks perl at least)
+LIBC[unexport] = "1"

-- 
Tom Rini
Mentor Graphics Corporation

Re: 'LIBC' variable clashes with perl variable

[Roman I Khimov]

[-- Attachment #1: Type: Text/Plain, Size: 537 bytes --]

В сообщении от Четверг 29 июля 2010 05:31:07 автор Tom Rini написал:
> bitbake.conf: Unexport LIBC from the environment
> 
> As spoted by Koen at least perl makes use of LIBC coming from the
> environment to be meaningful, so don't export it as it doesn't match our
> meaning.
> 
> Signed-off-by: Tom Rini <tom_rini@mentor.com>

Acked-by: Roman I Khimov <khimov@altell.ru>

-- 
 http://roman.khimov.ru
mailto: roman@khimov.ru
gpg --keyserver hkp://subkeys.pgp.net --recv-keys 0xE5E055C3

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 205 bytes --]