Patchwork Signed-off-by: Andrey Belous <abelous@broadcom.com>

login
register
mail settings
Submitter Andrey Belous
Date March 4, 2014, 8:40 p.m.
Message ID <1393965644-17393-1-git-send-email-abelous@broadcom.com>
Download mbox | patch
Permalink /patch/67979/
State New
Headers show

Comments

Andrey Belous - March 4, 2014, 8:40 p.m.
Setting of a variable PREFERRED_PROVIDER_virtual/libc only if it doesn't have a value
---
 meta/conf/distro/include/tclibc-uclibc.inc | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
Phil Blundell - March 4, 2014, 9:43 p.m.
On Tue, 2014-03-04 at 15:40 -0500, Andrey Belous wrote:
> -PREFERRED_PROVIDER_virtual/libc = "uclibc"
> +PREFERRED_PROVIDER_virtual/libc ?= "uclibc"

Why would you be including tclibc-uclibc.inc if you don't want to use
uclibc?

p.
Andrey Belous - March 4, 2014, 9:55 p.m.
Hi Phil,
I have prebuild uclibc library that I would like to use, without that change I cannot force to use my prebuild library

-Andrey

-----Original Message-----
From: Phil Blundell [mailto:pb@pbcl.net] 
Sent: Tuesday, March 04, 2014 4:43 PM
To: Andrey Belous
Cc: openembedded-core@lists.openembedded.org
Subject: Re: [OE-core] [PATCH] Signed-off-by: Andrey Belous <abelous@broadcom.com>

On Tue, 2014-03-04 at 15:40 -0500, Andrey Belous wrote:
> -PREFERRED_PROVIDER_virtual/libc = "uclibc"
> +PREFERRED_PROVIDER_virtual/libc ?= "uclibc"

Why would you be including tclibc-uclibc.inc if you don't want to use
uclibc?

p.
Saul Wold - March 4, 2014, 11:05 p.m.
On 03/04/2014 12:40 PM, Andrey Belous wrote:
> Setting of a variable PREFERRED_PROVIDER_virtual/libc only if it doesn't have a value

The format of your commit message is backwards, you need to have a short 
message with the filename:

tclibc-uclibc: <short message>

<long message>

Signed-off-by: ....


Also this kind of change is distro specific and could be in your distro 
specific information, don;'t you need to override more of what's in here 
if you are using a prebuilt library?

Sau!

> ---
>   meta/conf/distro/include/tclibc-uclibc.inc | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/meta/conf/distro/include/tclibc-uclibc.inc b/meta/conf/distro/include/tclibc-uclibc.inc
> index 9245112..a4d8c39 100644
> --- a/meta/conf/distro/include/tclibc-uclibc.inc
> +++ b/meta/conf/distro/include/tclibc-uclibc.inc
> @@ -8,7 +8,7 @@ LIBCEXTENSION = "-uclibc"
>   LIBCOVERRIDE = ":libc-uclibc"
>   OVERRIDES .= "${LIBCOVERRIDE}"
>
> -PREFERRED_PROVIDER_virtual/libc = "uclibc"
> +PREFERRED_PROVIDER_virtual/libc ?= "uclibc"
>   PREFERRED_PROVIDER_virtual/libiconv ?= "libiconv"
>   PREFERRED_PROVIDER_virtual/libintl ?= "gettext"
>   PREFERRED_PROVIDER_virtual/nativesdk-libintl ?= "nativesdk-eglibc"
>
Andrey Belous - March 4, 2014, 11:11 p.m.
Yes I overwriting more  , but other variable already with ?= and allow to override them

-Andrey 

-----Original Message-----
From: Saul Wold [mailto:sgw@linux.intel.com] 
Sent: Tuesday, March 04, 2014 6:05 PM
To: Andrey Belous; openembedded-core@lists.openembedded.org
Subject: Re: [OE-core] [PATCH] Signed-off-by: Andrey Belous <abelous@broadcom.com>

On 03/04/2014 12:40 PM, Andrey Belous wrote:
> Setting of a variable PREFERRED_PROVIDER_virtual/libc only if it doesn't have a value

The format of your commit message is backwards, you need to have a short 
message with the filename:

tclibc-uclibc: <short message>

<long message>

Signed-off-by: ....


Also this kind of change is distro specific and could be in your distro 
specific information, don;'t you need to override more of what's in here 
if you are using a prebuilt library?

Sau!

> ---
>   meta/conf/distro/include/tclibc-uclibc.inc | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/meta/conf/distro/include/tclibc-uclibc.inc b/meta/conf/distro/include/tclibc-uclibc.inc
> index 9245112..a4d8c39 100644
> --- a/meta/conf/distro/include/tclibc-uclibc.inc
> +++ b/meta/conf/distro/include/tclibc-uclibc.inc
> @@ -8,7 +8,7 @@ LIBCEXTENSION = "-uclibc"
>   LIBCOVERRIDE = ":libc-uclibc"
>   OVERRIDES .= "${LIBCOVERRIDE}"
>
> -PREFERRED_PROVIDER_virtual/libc = "uclibc"
> +PREFERRED_PROVIDER_virtual/libc ?= "uclibc"
>   PREFERRED_PROVIDER_virtual/libiconv ?= "libiconv"
>   PREFERRED_PROVIDER_virtual/libintl ?= "gettext"
>   PREFERRED_PROVIDER_virtual/nativesdk-libintl ?= "nativesdk-eglibc"
>
Nicolas Dechesne - March 6, 2014, 5:25 p.m.
Hi,

On Wed, Mar 5, 2014 at 7:11 AM, Andrey Belous <abelous@broadcom.com> wrote:
> Yes I overwriting more  , but other variable already with ?= and allow to override them

I think I am okay with the patch proposed here. i don't see why the
late assignment is missing just for this specific variable while it's
used for every other variables and in the eglibc include file we
already use the ?= for virtual/libc.

as far as I am concerned, assuming the commit log gets rewritten
properly, you can add:

Signed-off-by: Nicolas Dechesne <nicolas.dechesne@linaro.org>

cheers,

nico
Khem Raj - March 6, 2014, 10:43 p.m.
On Tue, Mar 4, 2014 at 1:55 PM, Andrey Belous <abelous@broadcom.com> wrote:
> Hi Phil,
> I have prebuild uclibc library that I would like to use, without that change I cannot force to use my prebuild library
>

OK so write a uclibc recipe for it and let it provide virtual/libc
and, we can very well have multiple version of recipes, this is no
different. Then use higher DEFAULT_PREFERENCE for your recipe.

> -Andrey
>
> -----Original Message-----
> From: Phil Blundell [mailto:pb@pbcl.net]
> Sent: Tuesday, March 04, 2014 4:43 PM
> To: Andrey Belous
> Cc: openembedded-core@lists.openembedded.org
> Subject: Re: [OE-core] [PATCH] Signed-off-by: Andrey Belous <abelous@broadcom.com>
>
> On Tue, 2014-03-04 at 15:40 -0500, Andrey Belous wrote:
>> -PREFERRED_PROVIDER_virtual/libc = "uclibc"
>> +PREFERRED_PROVIDER_virtual/libc ?= "uclibc"
>
> Why would you be including tclibc-uclibc.inc if you don't want to use
> uclibc?
>
> p.
>
>
> _______________________________________________
> Openembedded-core mailing list
> Openembedded-core@lists.openembedded.org
> http://lists.openembedded.org/mailman/listinfo/openembedded-core
Nicolas Dechesne - March 7, 2014, 10:33 a.m.
Hi Khem,

On Fri, Mar 7, 2014 at 6:43 AM, Khem Raj <raj.khem@gmail.com> wrote:
> On Tue, Mar 4, 2014 at 1:55 PM, Andrey Belous <abelous@broadcom.com> wrote:
>> Hi Phil,
>> I have prebuild uclibc library that I would like to use, without that change I cannot force to use my prebuild library
>>
>
> OK so write a uclibc recipe for it and let it provide virtual/libc
> and, we can very well have multiple version of recipes, this is no
> different. Then use higher DEFAULT_PREFERENCE for your recipe.

I still don't get it why in distro/include/tclibc-uclibc.inc we have

PREFERRED_PROVIDER_virtual/libc = "uclibc"
PREFERRED_PROVIDER_virtual/libiconv ?= "libiconv"
PREFERRED_PROVIDER_virtual/libintl ?= "gettext"
PREFERRED_PROVIDER_virtual/nativesdk-libintl ?= "nativesdk-eglibc"
PREFERRED_PROVIDER_virtual/nativesdk-libiconv ?= "nativesdk-eglibc"

and in distro/include/tclibc-eglibc.inc, we have:

PREFERRED_PROVIDER_virtual/libiconv ?= "eglibc"
PREFERRED_PROVIDER_virtual/nativesdk-libiconv ?= "nativesdk-eglibc"
PREFERRED_PROVIDER_virtual/nativesdk-libintl ?= "nativesdk-eglibc"
PREFERRED_PROVIDER_virtual/libintl ?= "eglibc"
PREFERRED_PROVIDER_virtual/libc ?= "eglibc"
PREFERRED_PROVIDER_virtual/nativesdk-libc ?= "nativesdk-eglibc"
PREFERRED_PROVIDER_virtual/libc-locale ?= "eglibc-locale"

Is a specific reason why in the first case we use "=" and not "?="
like used everywhere else? This file is loaded last, so we have no
chance to override the value. regardless of the actual reason, don't
we need the change anyways?
Phil Blundell - March 7, 2014, 6:26 p.m.
Fair enough.  I also agree that these variables should at least be
consistent, and if we're assigning everything else with ?= then there
seems no reason for that one provider to be treated specially.  So I
think your patch is fine, though it might be a good idea to re-send it
with a less messed up commit message if you didn't already do that.

p.

On Tue, 2014-03-04 at 21:55 +0000, Andrey Belous wrote:
> Hi Phil,
> I have prebuild uclibc library that I would like to use, without that change I cannot force to use my prebuild library
> 
> -Andrey
> 
> -----Original Message-----
> From: Phil Blundell [mailto:pb@pbcl.net] 
> Sent: Tuesday, March 04, 2014 4:43 PM
> To: Andrey Belous
> Cc: openembedded-core@lists.openembedded.org
> Subject: Re: [OE-core] [PATCH] Signed-off-by: Andrey Belous <abelous@broadcom.com>
> 
> On Tue, 2014-03-04 at 15:40 -0500, Andrey Belous wrote:
> > -PREFERRED_PROVIDER_virtual/libc = "uclibc"
> > +PREFERRED_PROVIDER_virtual/libc ?= "uclibc"
> 
> Why would you be including tclibc-uclibc.inc if you don't want to use
> uclibc?
> 
> p.
> 
>
Khem Raj - March 7, 2014, 6:40 p.m.
On Fri, Mar 7, 2014 at 2:33 AM, Nicolas Dechesne
<nicolas.dechesne@linaro.org> wrote:
> Hi Khem,
>
> On Fri, Mar 7, 2014 at 6:43 AM, Khem Raj <raj.khem@gmail.com> wrote:
>> On Tue, Mar 4, 2014 at 1:55 PM, Andrey Belous <abelous@broadcom.com> wrote:
>>> Hi Phil,
>>> I have prebuild uclibc library that I would like to use, without that change I cannot force to use my prebuild library
>>>
>>
>> OK so write a uclibc recipe for it and let it provide virtual/libc
>> and, we can very well have multiple version of recipes, this is no
>> different. Then use higher DEFAULT_PREFERENCE for your recipe.
>
> I still don't get it why in distro/include/tclibc-uclibc.inc we have
>
> PREFERRED_PROVIDER_virtual/libc = "uclibc"
> PREFERRED_PROVIDER_virtual/libiconv ?= "libiconv"
> PREFERRED_PROVIDER_virtual/libintl ?= "gettext"
> PREFERRED_PROVIDER_virtual/nativesdk-libintl ?= "nativesdk-eglibc"
> PREFERRED_PROVIDER_virtual/nativesdk-libiconv ?= "nativesdk-eglibc"
>
> and in distro/include/tclibc-eglibc.inc, we have:
>
> PREFERRED_PROVIDER_virtual/libiconv ?= "eglibc"
> PREFERRED_PROVIDER_virtual/nativesdk-libiconv ?= "nativesdk-eglibc"
> PREFERRED_PROVIDER_virtual/nativesdk-libintl ?= "nativesdk-eglibc"
> PREFERRED_PROVIDER_virtual/libintl ?= "eglibc"
> PREFERRED_PROVIDER_virtual/libc ?= "eglibc"
> PREFERRED_PROVIDER_virtual/nativesdk-libc ?= "nativesdk-eglibc"
> PREFERRED_PROVIDER_virtual/libc-locale ?= "eglibc-locale"
>
> Is a specific reason why in the first case we use "=" and not "?="
> like used everywhere else? This file is loaded last, so we have no
> chance to override the value. regardless of the actual reason, don't
> we need the change anyways?

I would think eglibc.inc should be fixed as well.

Patch

diff --git a/meta/conf/distro/include/tclibc-uclibc.inc b/meta/conf/distro/include/tclibc-uclibc.inc
index 9245112..a4d8c39 100644
--- a/meta/conf/distro/include/tclibc-uclibc.inc
+++ b/meta/conf/distro/include/tclibc-uclibc.inc
@@ -8,7 +8,7 @@  LIBCEXTENSION = "-uclibc"
 LIBCOVERRIDE = ":libc-uclibc"
 OVERRIDES .= "${LIBCOVERRIDE}"
 
-PREFERRED_PROVIDER_virtual/libc = "uclibc"
+PREFERRED_PROVIDER_virtual/libc ?= "uclibc"
 PREFERRED_PROVIDER_virtual/libiconv ?= "libiconv"
 PREFERRED_PROVIDER_virtual/libintl ?= "gettext"
 PREFERRED_PROVIDER_virtual/nativesdk-libintl ?= "nativesdk-eglibc"