RFC: feature request - wrap sources

RFC: feature request - wrap sources

[Robert Schuster]

[-- Attachment #1: Type: text/plain, Size: 593 bytes --]

Hi all,
I have a feature request that should help dealing with those source
tarballs which do not contain a proper top level directory. The idea is
to allow to add a parameter 'wrap' to the SRC_URI entry denoting the
source tarball. If it is present (and 1) the unpacker creates a
directory named ${P} and unpacks the tarball therein.

This would make it unneccessary to change S for such packages and
furthermore the work directory is not cluttered with files and folders
from inside the source tarball.

Like it? (Any pointer where such code should be put?)

Regards
Robert


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 260 bytes --]

Re: RFC: feature request - wrap sources

[Richard Purdie]

On Fri, 2007-12-21 at 09:34 +0100, Robert Schuster wrote:
> I have a feature request that should help dealing with those source
> tarballs which do not contain a proper top level directory. The idea is
> to allow to add a parameter 'wrap' to the SRC_URI entry denoting the
> source tarball. If it is present (and 1) the unpacker creates a
> directory named ${P} and unpacks the tarball therein.
> 
> This would make it unneccessary to change S for such packages and
> furthermore the work directory is not cluttered with files and folders
> from inside the source tarball.
> 
> Like it? (Any pointer where such code should be put?)

Would it not be simpler just to specify an attribute like unpacksubdir
which you could set to ${PN}-${PV}?

That way you have a generic and simpler solution?

Cheers,

Richard

Re: RFC: feature request - wrap sources

[Robert Schuster]

[-- Attachment #1: Type: text/plain, Size: 1125 bytes --]

Richard Purdie schrieb:
> On Fri, 2007-12-21 at 09:34 +0100, Robert Schuster wrote:
>> I have a feature request that should help dealing with those source
>> tarballs which do not contain a proper top level directory. The idea is
>> to allow to add a parameter 'wrap' to the SRC_URI entry denoting the
>> source tarball. If it is present (and 1) the unpacker creates a
>> directory named ${P} and unpacks the tarball therein.
>>
>> This would make it unneccessary to change S for such packages and
>> furthermore the work directory is not cluttered with files and folders
>> from inside the source tarball.
>>
>> Like it? (Any pointer where such code should be put?)
> 
> Would it not be simpler just to specify an attribute like unpacksubdir
> which you could set to ${PN}-${PV}?
Specifying the directory to be created would be nice. However one of my
idea was that by using such a mechanism S would be correct.

Would it be possible to automatically set S to ${WORKDIR}/<anything> if
unpacksubdir=<anything> is used? If so we could have genericity and
saving boilerplate code. :)

Regards
Robert


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 260 bytes --]

Re: RFC: feature request - wrap sources

[Paul Sokolovsky]

Hello Robert,

Friday, December 21, 2007, 12:24:48 PM, you wrote:

> Richard Purdie schrieb:
>> On Fri, 2007-12-21 at 09:34 +0100, Robert Schuster wrote:
>>> I have a feature request that should help dealing with those source
>>> tarballs which do not contain a proper top level directory. The idea is
>>> to allow to add a parameter 'wrap' to the SRC_URI entry denoting the
>>> source tarball. If it is present (and 1) the unpacker creates a
>>> directory named ${P} and unpacks the tarball therein.
>>>
>>> This would make it unneccessary to change S for such packages and
>>> furthermore the work directory is not cluttered with files and folders
>>> from inside the source tarball.
>>>
>>> Like it? (Any pointer where such code should be put?)
>> 
>> Would it not be simpler just to specify an attribute like unpacksubdir
>> which you could set to ${PN}-${PV}?
> Specifying the directory to be created would be nice. However one of my
> idea was that by using such a mechanism S would be correct.

> Would it be possible to automatically set S to ${WORKDIR}/<anything> if
> unpacksubdir=<anything> is used? If so we could have genericity and
> saving boilerplate code. :)

  Why, instead, S has default value of "${WORKDIR}/${P}", so just fit
unpacksubdir up to that. And that unpacksubdir is yet stuff to implement
AFAIU.

> Regards
> Robert




-- 
Best regards,
 Paul                            mailto:pmiscml@gmail.com

Re: RFC: feature request - wrap sources

[Richard Purdie]

On Fri, 2007-12-21 at 13:16 +0200, Paul Sokolovsky wrote:
> Friday, December 21, 2007, 12:24:48 PM, you wrote:
> > Richard Purdie schrieb:
> >> On Fri, 2007-12-21 at 09:34 +0100, Robert Schuster wrote:
> >> Would it not be simpler just to specify an attribute like unpacksubdir
> >> which you could set to ${PN}-${PV}?
> > Specifying the directory to be created would be nice. However one of my
> > idea was that by using such a mechanism S would be correct.
> 
> > Would it be possible to automatically set S to ${WORKDIR}/<anything> if
> > unpacksubdir=<anything> is used? If so we could have genericity and
> > saving boilerplate code. :)
> 
>   Why, instead, S has default value of "${WORKDIR}/${P}", so just fit
> unpacksubdir up to that. And that unpacksubdir is yet stuff to implement
> AFAIU.

This was my intent although I perhaps didn't make it very clear. If you
specify "unpacksubdir=${P}" (P = ${PN}-${PV}) the default for WORKDIR
would just work and wouldn't need to be changed.

Cheers,

Richard