[PATCH 0/1] native.bbclass: remove invalid CONFIG_SITE

[PATCH 0/1] native.bbclass: remove invalid CONFIG_SITE

[Robert Yang]

The following changes since commit a4edfa4cf451bf412525887b5b24b9db6486ae97:

  remove unused distutils-tools.bbclass (2019-01-23 07:57:02 +0000)

are available in the git repository at:

  git://git.openembedded.org/openembedded-core-contrib rbt/native
  http://cgit.openembedded.org/openembedded-core-contrib/log/?h=rbt/native

Robert Yang (1):
  native.bbclass: remove invalid CONFIG_SITE

 meta/classes/native.bbclass | 3 ---
 meta/site/native            | 1 -
 2 files changed, 4 deletions(-)
 delete mode 100644 meta/site/native

-- 
2.7.4

[PATCH 1/1] native.bbclass: remove invalid CONFIG_SITE

[Robert Yang]

This CONFIG_SITE has no effect since autotools.bbclass handles it. And the
comment line is out of date, it was for "CONFIG_SITE = ''", so remove them.

Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
---
 meta/classes/native.bbclass | 3 ---
 meta/site/native            | 1 -
 2 files changed, 4 deletions(-)
 delete mode 100644 meta/site/native

diff --git a/meta/classes/native.bbclass b/meta/classes/native.bbclass
index ddccfe2..c20c6dc 100644
--- a/meta/classes/native.bbclass
+++ b/meta/classes/native.bbclass
@@ -54,9 +54,6 @@ TOOLCHAIN_OPTIONS = ""
 # Don't build ptest natively
 PTEST_ENABLED = "0"
 
-# Don't use site files for native builds
-export CONFIG_SITE = "${COREBASE}/meta/site/native"
-
 # set the compiler as well. It could have been set to something else
 export CC = "${BUILD_CC}"
 export CXX = "${BUILD_CXX}"
diff --git a/meta/site/native b/meta/site/native
deleted file mode 100644
index 7dfb1cb..0000000
--- a/meta/site/native
+++ /dev/null
@@ -1 +0,0 @@
-ac_cv_path_SED=sed
-- 
2.7.4

Re: [PATCH 1/1] native.bbclass: remove invalid CONFIG_SITE

[Richard Purdie]

On Thu, 2019-01-24 at 15:06 +0800, Robert Yang wrote:
> This CONFIG_SITE has no effect since autotools.bbclass handles it.
> And the
> comment line is out of date, it was for "CONFIG_SITE = ''", so remove
> them.

Going from memory, we had a problem where the path to sed was being
encoded into scripts (e.g. libtool/autotools and others). It may be
/usr/bin/sed on some systems and /bin/sed on others.

Are you sure this is no longer a problem?

Cheers,

Richard

> Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
> ---
>  meta/classes/native.bbclass | 3 ---
>  meta/site/native            | 1 -
>  2 files changed, 4 deletions(-)
>  delete mode 100644 meta/site/native
> 
> diff --git a/meta/classes/native.bbclass
> b/meta/classes/native.bbclass
> index ddccfe2..c20c6dc 100644
> --- a/meta/classes/native.bbclass
> +++ b/meta/classes/native.bbclass
> @@ -54,9 +54,6 @@ TOOLCHAIN_OPTIONS = ""
>  # Don't build ptest natively
>  PTEST_ENABLED = "0"
>  
> -# Don't use site files for native builds
> -export CONFIG_SITE = "${COREBASE}/meta/site/native"
> -
>  # set the compiler as well. It could have been set to something else
>  export CC = "${BUILD_CC}"
>  export CXX = "${BUILD_CXX}"
> diff --git a/meta/site/native b/meta/site/native
> deleted file mode 100644
> index 7dfb1cb..0000000
> --- a/meta/site/native
> +++ /dev/null
> @@ -1 +0,0 @@
> -ac_cv_path_SED=sed
> -- 
> 2.7.4
> 

Re: [PATCH 1/1] native.bbclass: remove invalid CONFIG_SITE

[Robert Yang]

On 1/24/19 8:51 PM, Richard Purdie wrote:
> On Thu, 2019-01-24 at 15:06 +0800, Robert Yang wrote:
>> This CONFIG_SITE has no effect since autotools.bbclass handles it.
>> And the
>> comment line is out of date, it was for "CONFIG_SITE = ''", so remove
>> them.
> 
> Going from memory, we had a problem where the path to sed was being
> encoded into scripts (e.g. libtool/autotools and others). It may be
> /usr/bin/sed on some systems and /bin/sed on others.
> 
> Are you sure this is no longer a problem?

Yes, that might be a problem, please ignore the patch, and I will update the
comment:

"# Don't use site files for native builds"

To:

# Only use site/native for native builds

// Robert

> 
> Cheers,
> 
> Richard
> 
>> Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
>> ---
>>   meta/classes/native.bbclass | 3 ---
>>   meta/site/native            | 1 -
>>   2 files changed, 4 deletions(-)
>>   delete mode 100644 meta/site/native
>>
>> diff --git a/meta/classes/native.bbclass
>> b/meta/classes/native.bbclass
>> index ddccfe2..c20c6dc 100644
>> --- a/meta/classes/native.bbclass
>> +++ b/meta/classes/native.bbclass
>> @@ -54,9 +54,6 @@ TOOLCHAIN_OPTIONS = ""
>>   # Don't build ptest natively
>>   PTEST_ENABLED = "0"
>>   
>> -# Don't use site files for native builds
>> -export CONFIG_SITE = "${COREBASE}/meta/site/native"
>> -
>>   # set the compiler as well. It could have been set to something else
>>   export CC = "${BUILD_CC}"
>>   export CXX = "${BUILD_CXX}"
>> diff --git a/meta/site/native b/meta/site/native
>> deleted file mode 100644
>> index 7dfb1cb..0000000
>> --- a/meta/site/native
>> +++ /dev/null
>> @@ -1 +0,0 @@
>> -ac_cv_path_SED=sed
>> -- 
>> 2.7.4
>>
> 
> 

Re: [PATCH 1/1] native.bbclass: remove invalid CONFIG_SITE

[Robert Yang]

On 1/25/19 10:40 AM, Robert Yang wrote:
> 
> 
> On 1/24/19 8:51 PM, Richard Purdie wrote:
>> On Thu, 2019-01-24 at 15:06 +0800, Robert Yang wrote:
>>> This CONFIG_SITE has no effect since autotools.bbclass handles it.
>>> And the
>>> comment line is out of date, it was for "CONFIG_SITE = ''", so remove
>>> them.
>>
>> Going from memory, we had a problem where the path to sed was being
>> encoded into scripts (e.g. libtool/autotools and others). It may be
>> /usr/bin/sed on some systems and /bin/sed on others.
>>
>> Are you sure this is no longer a problem?
> 
> Yes, that might be a problem, please ignore the patch, and I will update the
> comment:

After more investigations, I'm sure it is not longer needed, I was mis-leaded
by:

$ bitbake m4-native -cconfigure -e > m4-native.py

It shows that CONFIG_SITE="/buildarea1/lyang1/poky/meta/site/native"

But if I run:
$ bitbake m4-native -cconfigure

And check temp/run.do_configure:
CONFIG_SITE="/buildarea1/lyang1/poky/meta/site/endian-little 
/buildarea1/lyang1/poky/meta/site/common-linux 
/buildarea1/lyang1/poky/meta/site/common-glibc 
/buildarea1/lyang1/poky/meta/site/x86_64-linux 
/buildarea1/lyang1/poky/meta/site/common "

So the site/native was never used, and I think that it can be removed safely.

// Robert

> 
> "# Don't use site files for native builds"
> 
> To:
> 
> # Only use site/native for native builds
> 
> // Robert
> 
>>
>> Cheers,
>>
>> Richard
>>
>>> Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
>>> ---
>>>   meta/classes/native.bbclass | 3 ---
>>>   meta/site/native            | 1 -
>>>   2 files changed, 4 deletions(-)
>>>   delete mode 100644 meta/site/native
>>>
>>> diff --git a/meta/classes/native.bbclass
>>> b/meta/classes/native.bbclass
>>> index ddccfe2..c20c6dc 100644
>>> --- a/meta/classes/native.bbclass
>>> +++ b/meta/classes/native.bbclass
>>> @@ -54,9 +54,6 @@ TOOLCHAIN_OPTIONS = ""
>>>   # Don't build ptest natively
>>>   PTEST_ENABLED = "0"
>>> -# Don't use site files for native builds
>>> -export CONFIG_SITE = "${COREBASE}/meta/site/native"
>>> -
>>>   # set the compiler as well. It could have been set to something else
>>>   export CC = "${BUILD_CC}"
>>>   export CXX = "${BUILD_CXX}"
>>> diff --git a/meta/site/native b/meta/site/native
>>> deleted file mode 100644
>>> index 7dfb1cb..0000000
>>> --- a/meta/site/native
>>> +++ /dev/null
>>> @@ -1 +0,0 @@
>>> -ac_cv_path_SED=sed
>>> -- 
>>> 2.7.4
>>>
>>
>>

Re: [PATCH 1/1] native.bbclass: remove invalid CONFIG_SITE

[Robert Yang]

On 1/25/19 10:40 AM, Robert Yang wrote:
> 
> 
> On 1/24/19 8:51 PM, Richard Purdie wrote:
>> On Thu, 2019-01-24 at 15:06 +0800, Robert Yang wrote:
>>> This CONFIG_SITE has no effect since autotools.bbclass handles it.
>>> And the
>>> comment line is out of date, it was for "CONFIG_SITE = ''", so remove
>>> them.
>>
>> Going from memory, we had a problem where the path to sed was being
>> encoded into scripts (e.g. libtool/autotools and others). It may be
>> /usr/bin/sed on some systems and /bin/sed on others.
>>
>> Are you sure this is no longer a problem?
> 
> Yes, that might be a problem, please ignore the patch, and I will update the
> comment:

After more investigations, I'm sure that site/native can be removed safely,
I was mis-leaded by the following command just now:

$ bitbake m4-native -cconfigure -e > m4-native.py

It shows that CONFIG_SITE="/buildarea1/lyang1/poky/meta/site/native"

But if I run:
$ bitbake m4-native -cconfigure

And check temp/run.do_configure:
CONFIG_SITE="/buildarea1/lyang1/poky/meta/site/endian-little 
/buildarea1/lyang1/poky/meta/site/common-linux 
/buildarea1/lyang1/poky/meta/site/common-glibc 
/buildarea1/lyang1/poky/meta/site/x86_64-linux 
/buildarea1/lyang1/poky/meta/site/common "

So the site/native was never used, and I think that it can be removed safely.

The "$ bitbake m4-native -cconfigure -e" showed incorrect CONFIG_SITE, is this
a bug, please ?

// Robert

> 
> "# Don't use site files for native builds"
> 
> To:
> 
> # Only use site/native for native builds
> 
> // Robert
> 
>>
>> Cheers,
>>
>> Richard
>>
>>> Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
>>> ---
>>>   meta/classes/native.bbclass | 3 ---
>>>   meta/site/native            | 1 -
>>>   2 files changed, 4 deletions(-)
>>>   delete mode 100644 meta/site/native
>>>
>>> diff --git a/meta/classes/native.bbclass
>>> b/meta/classes/native.bbclass
>>> index ddccfe2..c20c6dc 100644
>>> --- a/meta/classes/native.bbclass
>>> +++ b/meta/classes/native.bbclass
>>> @@ -54,9 +54,6 @@ TOOLCHAIN_OPTIONS = ""
>>>   # Don't build ptest natively
>>>   PTEST_ENABLED = "0"
>>> -# Don't use site files for native builds
>>> -export CONFIG_SITE = "${COREBASE}/meta/site/native"
>>> -
>>>   # set the compiler as well. It could have been set to something else
>>>   export CC = "${BUILD_CC}"
>>>   export CXX = "${BUILD_CXX}"
>>> diff --git a/meta/site/native b/meta/site/native
>>> deleted file mode 100644
>>> index 7dfb1cb..0000000
>>> --- a/meta/site/native
>>> +++ /dev/null
>>> @@ -1 +0,0 @@
>>> -ac_cv_path_SED=sed
>>> -- 
>>> 2.7.4
>>>
>>
>>

Re: [PATCH 1/1] native.bbclass: remove invalid CONFIG_SITE

[Robert Yang]

On 1/25/19 10:59 AM, Robert Yang wrote:
> 
> 
> On 1/25/19 10:40 AM, Robert Yang wrote:
>>
>>
>> On 1/24/19 8:51 PM, Richard Purdie wrote:
>>> On Thu, 2019-01-24 at 15:06 +0800, Robert Yang wrote:
>>>> This CONFIG_SITE has no effect since autotools.bbclass handles it.
>>>> And the
>>>> comment line is out of date, it was for "CONFIG_SITE = ''", so remove
>>>> them.
>>>
>>> Going from memory, we had a problem where the path to sed was being
>>> encoded into scripts (e.g. libtool/autotools and others). It may be
>>> /usr/bin/sed on some systems and /bin/sed on others.

And for the sed issue, it has been set in meta/site/common:

ac_cv_path_SED=${ac_cv_path_SED=sed}

// Robert

>>>
>>> Are you sure this is no longer a problem?
>>
>> Yes, that might be a problem, please ignore the patch, and I will update the
>> comment:
> 
> After more investigations, I'm sure that site/native can be removed safely,
> I was mis-leaded by the following command just now:
> 
> $ bitbake m4-native -cconfigure -e > m4-native.py
> 
> It shows that CONFIG_SITE="/buildarea1/lyang1/poky/meta/site/native"
> 
> But if I run:
> $ bitbake m4-native -cconfigure
> 
> And check temp/run.do_configure:
> CONFIG_SITE="/buildarea1/lyang1/poky/meta/site/endian-little 
> /buildarea1/lyang1/poky/meta/site/common-linux 
> /buildarea1/lyang1/poky/meta/site/common-glibc 
> /buildarea1/lyang1/poky/meta/site/x86_64-linux 
> /buildarea1/lyang1/poky/meta/site/common "
> 
> So the site/native was never used, and I think that it can be removed safely.
> 
> The "$ bitbake m4-native -cconfigure -e" showed incorrect CONFIG_SITE, is this
> a bug, please ?
> 
> // Robert
> 
>>
>> "# Don't use site files for native builds"
>>
>> To:
>>
>> # Only use site/native for native builds
>>
>> // Robert
>>
>>>
>>> Cheers,
>>>
>>> Richard
>>>
>>>> Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
>>>> ---
>>>>   meta/classes/native.bbclass | 3 ---
>>>>   meta/site/native            | 1 -
>>>>   2 files changed, 4 deletions(-)
>>>>   delete mode 100644 meta/site/native
>>>>
>>>> diff --git a/meta/classes/native.bbclass
>>>> b/meta/classes/native.bbclass
>>>> index ddccfe2..c20c6dc 100644
>>>> --- a/meta/classes/native.bbclass
>>>> +++ b/meta/classes/native.bbclass
>>>> @@ -54,9 +54,6 @@ TOOLCHAIN_OPTIONS = ""
>>>>   # Don't build ptest natively
>>>>   PTEST_ENABLED = "0"
>>>> -# Don't use site files for native builds
>>>> -export CONFIG_SITE = "${COREBASE}/meta/site/native"
>>>> -
>>>>   # set the compiler as well. It could have been set to something else
>>>>   export CC = "${BUILD_CC}"
>>>>   export CXX = "${BUILD_CXX}"
>>>> diff --git a/meta/site/native b/meta/site/native
>>>> deleted file mode 100644
>>>> index 7dfb1cb..0000000
>>>> --- a/meta/site/native
>>>> +++ /dev/null
>>>> @@ -1 +0,0 @@
>>>> -ac_cv_path_SED=sed
>>>> -- 
>>>> 2.7.4
>>>>
>>>
>>>