Re: [gentoo-dev] [PATCH 6/6] Generate python depstrings in python-r1.

[Ian Stakenvicius]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

On 19/09/12 01:37 PM, Michał Górny wrote:
> On Wed, 19 Sep 2012 09:00:49 -0400 Ian Stakenvicius
> <axs@g.o> wrote:
>
>> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256
>>
>> On 19/09/12 06:00 AM, Micha? Górny wrote:
>>> --- gx86/eclass/python-distutils-ng.eclass | 20
>>> ++------------------ gx86/eclass/python-r1.eclass           |
>>> 28 ++++++++++++++++++++++++++-- 2 files changed, 28
>>> insertions(+), 20 deletions(-)
>>>
>>> diff --git a/gx86/eclass/python-distutils-ng.eclass
>>> b/gx86/eclass/python-distutils-ng.eclass index
>>> 5df965c..bffe563 100644 ---
>>> a/gx86/eclass/python-distutils-ng.eclass +++
>>> b/gx86/eclass/python-distutils-ng.eclass @@ -74,24 +74,8 @@
>>> _python-distutils-ng_get_binary_for_implementation() { esac }
>>>
>>> -for impl in ${PYTHON_COMPAT[@]}; do -	dep_str="${impl/_/.}" -
>>> case "${dep_str}" in -		python?.?) -
>>> dep_str="dev-lang/python:${dep_str: -3}" ;; - jython?.?) -
>>> dep_str="dev-java/jython:${dep_str: -3}" ;; -		pypy?.?) -
>>> dep_str="dev-python/pypy:${dep_str: -3}" ;; -		*) -			die
>>> "Unsupported implementation: ${impl}" ;; -	esac -
>>> dep_str="python_targets_${impl}? ( ${dep_str} )" - -
>>> RDEPEND="${RDEPEND} ${dep_str}" -	DEPEND="${DEPEND} ${dep_str}"
>>> - unset dep_str -done +RDEPEND=${PYTHON_DEPEND}
>>> +DEPEND=${PYTHON_DEPEND}
>>>
>>> _PACKAGE_SPECIFIC_S="${S#${WORKDIR}/}"
>>>
>>> diff --git a/gx86/eclass/python-r1.eclass
>>> b/gx86/eclass/python-r1.eclass index 18f9246..3017dd8 100644
>>> --- a/gx86/eclass/python-r1.eclass +++
>>> b/gx86/eclass/python-r1.eclass @@ -40,14 +40,38 @@
>>> _PYTHON_ALL_IMPLS=( # a package supports. It must be set before
>>> the `inherit' call. # The default is to enable all
>>> implementations. # -# PYTHON_COMPAT can be either a scalar or
>>> an array. If it's a scalar, the eclass -# will implicitly
>>> convert it to an array. +# PYTHON_COMPAT can be either a scalar
>>> or an array. If it's a scalar, +# the eclass will implicitly
>>> convert it to an array. :
>>> ${PYTHON_COMPAT:=${_PYTHON_ALL_IMPLS[@]}}
>>>
>>> +# @ECLASS-VARIABLE: PYTHON_DEPEND +# @DESCRIPTION: +# This is
>>> an eclass-generated Python dependency string for all +#
>>> implementations listed in PYTHON_COMPAT. + PYTHON_COMPAT=(
>>> ${PYTHON_COMPAT[@]} )
>>>
>>> _python_set_globals() {
>>> IUSE=${PYTHON_COMPAT[@]/#/python_targets_} REQUIRED_USE="|| (
>>> ${IUSE} )" + +	PYTHON_DEPEND= + local i +	for i in
>>> ${PYTHON_COMPAT[@]}; do + local d +		case ${i} in + python*) +
>>> d='dev-lang/python';; +			jython*) + d='dev-java/jython';; +
>>> pypy*) + d='dev-python/pypy';; +			*) +				die "Invalid
>>> implementation: ${i}" +		esac + +		local v=${i##*[a-z]} +
>>> PYTHON_DEPEND+=" python_targets_${i}? ( ${d}:${v/_/.} )" + done
>>> } _python_set_globals
>>
>>
>> I think we really need to use a different variable than
>> "PYTHON_DEPEND".  There are ebuilds that inherit both
>> python.eclass and python-distutils-ng.eclass and so this will
>> obviously break those cases..  (that said, I do think that
>> python-r1.eclass should gain whatever bits it is missing so said
>> inherit on python.eclass isn't necessary, but that's beside the
>> point)
>
> I'm not sure if we shouldn't forbid inheriting both eclasses.
> That's a bit risky, at least. And it's very likely that someone is
> doing something very wrong and not noticing the failure yet.
>

We can forbid a double-inherit, but that doesn't mean we should allow
PYTHON_DEPEND to be overloaded with two different meanings.

I would recommend instead to use 'PYTHON_DEPS'; it's different, and is
a more generic name which would seem to fit better when appended to
either/both DEPEND and RDEPEND.

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.19 (GNU/Linux)

iF4EAREIAAYFAlBaB74ACgkQ2ugaI38ACPD4mwD/e++PURTyeBoeFZodadFVDBYh
tBIuTaDFeuR7+PKNgPwA/1WFz8urK1ErpDLWTVsLqIUZra0L0RFG5/d+8jr3DPVQ
=bZPZ
-----END PGP SIGNATURE-----