[1,of,2,v2] infra: introduce suitable-extractor helper function

In order to simplify determining the right extractor tool for a given
file type, this patch introduces a make function 'suitable-extractor'.
Its usage is $(call suitable-extractor,filename), and it returns the
path to the suitable extractor.

Signed-off-by: Thomas De Schampheleire <thomas.de.schampheleire@gmail.com>

---
(v2): new patch in series

Note: one remaining direct usage of the INFLATE variables is in
package/pkg-generic.mk, but it's in the inner block where dollar signs
are thrown around your ears. I'm not sure if it's possible to use the
function there...

 package/lsof/lsof.mk                     |  2 +-
 package/perl/perl.mk                     |  2 +-
 package/pkg-generic.mk                   |  2 +-
 package/pkg-utils.mk                     |  2 ++
 package/tar/tar.mk                       |  2 +-
 toolchain/toolchain-external/ext-tool.mk |  6 +++---
 6 files changed, 9 insertions(+), 7 deletions(-)

Dear Thomas De Schampheleire,

On Thu, 01 Aug 2013 22:55:47 +0200, Thomas De Schampheleire wrote:

> diff --git a/package/pkg-utils.mk b/package/pkg-utils.mk
> --- a/package/pkg-utils.mk
> +++ b/package/pkg-utils.mk
> @@ -62,6 +62,8 @@ INFLATE.tbz2 = $(BZCAT)
>  INFLATE.tgz  = $(ZCAT)
>  INFLATE.xz   = $(XZCAT)
>  INFLATE.tar  = cat
> +# suitable-extractor(filename): returns extractor based on suffix
> +suitable-extractor = $(firstword $(INFLATE$(suffix $(1))))

Do you know why we need this $(firstword ...) call here? In all places
in was using directly $(INFLATE$(...)), except in the package
infrastructure where it was doing this firstword additional call.

It was added by Peter in 2c6390a5d0c01420879e9f23bc89afb19976da4a.

Ideas?

Thomas

On Fri, Aug 2, 2013 at 10:28 AM, Thomas Petazzoni
<thomas.petazzoni@free-electrons.com> wrote:
> Dear Thomas De Schampheleire,
>
> On Thu, 01 Aug 2013 22:55:47 +0200, Thomas De Schampheleire wrote:
>
>> diff --git a/package/pkg-utils.mk b/package/pkg-utils.mk
>> --- a/package/pkg-utils.mk
>> +++ b/package/pkg-utils.mk
>> @@ -62,6 +62,8 @@ INFLATE.tbz2 = $(BZCAT)
>>  INFLATE.tgz  = $(ZCAT)
>>  INFLATE.xz   = $(XZCAT)
>>  INFLATE.tar  = cat
>> +# suitable-extractor(filename): returns extractor based on suffix
>> +suitable-extractor = $(firstword $(INFLATE$(suffix $(1))))
>
> Do you know why we need this $(firstword ...) call here? In all places
> in was using directly $(INFLATE$(...)), except in the package
> infrastructure where it was doing this firstword additional call.
>
> It was added by Peter in 2c6390a5d0c01420879e9f23bc89afb19976da4a.

It wasn't clear to me when I sent the patch, but I figured it didn't hurt.
But now I think I know: the inflate targets originate from the
.config, and users can add extra arguments there. In fact, the default
for $(ZCAT) is:
BR2_ZCAT="gzip -d -c"

To check the dependency, we only want to check for 'gzip', but to do
the actual inflate, we shouldn't use 'firstword'. So, the patch is
wrong, and I will fix it :)

Thanks for highlighting this!

Best regards,
Thomas

Dear Thomas De Schampheleire,

On Fri, 2 Aug 2013 10:50:44 +0200, Thomas De Schampheleire wrote:

> > Do you know why we need this $(firstword ...) call here? In all places
> > in was using directly $(INFLATE$(...)), except in the package
> > infrastructure where it was doing this firstword additional call.
> >
> > It was added by Peter in 2c6390a5d0c01420879e9f23bc89afb19976da4a.
> 
> It wasn't clear to me when I sent the patch, but I figured it didn't hurt.
> But now I think I know: the inflate targets originate from the
> .config, and users can add extra arguments there. In fact, the default
> for $(ZCAT) is:
> BR2_ZCAT="gzip -d -c"
> 
> To check the dependency, we only want to check for 'gzip', but to do
> the actual inflate, we shouldn't use 'firstword'. So, the patch is
> wrong, and I will fix it :)

Ok, makes sense to me now. Indeed this makes the patch wrong :)

Are you annoyed if I integrate this in -next ? I want to release -rc1
today, and I think this kind of stuff is a little bit too sensitive to
be merged now. From your quick feedback on those patches, I have the
feeling that you wanted it in 2013.08, so I don't want to ruin your
hope, but I believe it's a bit too late now. What do you think?

Best regards,

Thomas

Hi Thomas,

On Fri, Aug 2, 2013 at 10:56 AM, Thomas Petazzoni
<thomas.petazzoni@free-electrons.com> wrote:
> Dear Thomas De Schampheleire,
>
> On Fri, 2 Aug 2013 10:50:44 +0200, Thomas De Schampheleire wrote:
>
>> > Do you know why we need this $(firstword ...) call here? In all places
>> > in was using directly $(INFLATE$(...)), except in the package
>> > infrastructure where it was doing this firstword additional call.
>> >
>> > It was added by Peter in 2c6390a5d0c01420879e9f23bc89afb19976da4a.
>>
>> It wasn't clear to me when I sent the patch, but I figured it didn't hurt.
>> But now I think I know: the inflate targets originate from the
>> .config, and users can add extra arguments there. In fact, the default
>> for $(ZCAT) is:
>> BR2_ZCAT="gzip -d -c"
>>
>> To check the dependency, we only want to check for 'gzip', but to do
>> the actual inflate, we shouldn't use 'firstword'. So, the patch is
>> wrong, and I will fix it :)
>
> Ok, makes sense to me now. Indeed this makes the patch wrong :)
>
> Are you annoyed if I integrate this in -next ? I want to release -rc1
> today, and I think this kind of stuff is a little bit too sensitive to
> be merged now. From your quick feedback on those patches, I have the
> feeling that you wanted it in 2013.08, so I don't want to ruin your
> hope, but I believe it's a bit too late now. What do you think?
>

It's always nice if patches make it to the upcoming release and not
the one after that :)
But I understand that it's not critical and could have side-effects we
didn't think of, so no problem to put it in -next. I won't hate you
for it ;)

Best regards,
Thomas