gentoo-x86 commit in sci-geosciences/mapserver: mapserver-5.4.2-r1.ebuild ChangeLog
 

On 12/23/2010 12:18 PM, Jeremy Olexa wrote:
> FYI: This adds no value to the ebuild. The default src_unpack is ran
> when src_unpack() is not defined.

Unless src_unpack() is exported in one of the many eclasses that ebuild
inherits. In which case this would "un-export" it and bring the behavior
back to default.

gentoo-x86 commit in sci-geosciences/mapserver: mapserver-5.4.2-r1.ebuild ChangeLog
 

On 23/12/10 19:47, Mike Gilbert wrote:
> On 12/23/2010 12:18 PM, Jeremy Olexa wrote:
>> FYI: This adds no value to the ebuild. The default src_unpack is ran
>> when src_unpack() is not defined.
>
> Unless src_unpack() is exported in one of the many eclasses that ebuild
> inherits. In which case this would "un-export" it and bring the behavior
> back to default.
>

That's right. The magic comes from the loved python.eclass. If it is not
added the webapp.eclass does some things.

gentoo-x86 commit in sci-geosciences/mapserver: mapserver-5.4.2-r1.ebuild ChangeLog
 

On Thursday, December 23, 2010 14:07:45 Justin (jlec) Lecher wrote:
> On 23/12/10 19:47, Mike Gilbert wrote:
> > On 12/23/2010 12:18 PM, Jeremy Olexa wrote:
> >> FYI: This adds no value to the ebuild. The default src_unpack is ran
> >> when src_unpack() is not defined.
> >
> > Unless src_unpack() is exported in one of the many eclasses that ebuild
> > inherits. In which case this would "un-export" it and bring the behavior
> > back to default.
>
> That's right. The magic comes from the loved python.eclass. If it is not
> added the webapp.eclass does some things.

sounds like the ebuild should have:
# avoid inherited eclass stupidity
-mike

gentoo-x86 commit in sci-geosciences/mapserver: mapserver-5.4.2-r1.ebuild ChangeLog
 

2010-12-23 20:07:45 Justin (jlec) Lecher napisał(a):
> On 23/12/10 19:47, Mike Gilbert wrote:
> > On 12/23/2010 12:18 PM, Jeremy Olexa wrote:
> >> FYI: This adds no value to the ebuild. The default src_unpack is ran
> >> when src_unpack() is not defined.
> >
> > Unless src_unpack() is exported in one of the many eclasses that ebuild
> > inherits. In which case this would "un-export" it and bring the behavior
> > back to default.
> >
>
> That's right. The magic comes from the loved python.eclass. If it is not
> added the webapp.eclass does some things.

What do you mean about python.eclass?
python.eclass doesn't define python_src_unpack().

--
Arfrever Frehtes Taifersar Arahesis

gentoo-x86 commit in sci-geosciences/mapserver: mapserver-5.4.2-r1.ebuild ChangeLog
 

On 12/23/2010 01:07 PM, Justin (jlec) Lecher wrote:

On 23/12/10 19:47, Mike Gilbert wrote:

On 12/23/2010 12:18 PM, Jeremy Olexa wrote:

FYI: This adds no value to the ebuild. The default src_unpack is ran
when src_unpack() is not defined.


Unless src_unpack() is exported in one of the many eclasses that ebuild
inherits. In which case this would "un-export" it and bring the behavior
back to default.



That's right. The magic comes from the loved python.eclass. If it is not
added the webapp.eclass does some things.



You know...I did test the results before posting to the -dev list ;) So,
yes, I knew that there was a interesting call in src_unpack from the
ruby eclass, but it didn't change the results. You could add a comment
on WHY you explicitly define the function in the future, as I am
guessing no one besides you currently knows.. :)

-Jeremy

% diff -ur mapserver-5.4.2-r1.orig/ mapserver-5.4.2-r1.edit/
File mapserver-5.4.2-r1.orig/.ipc_in is a fifo while file
mapserver-5.4.2-r1.edit/.ipc_in is a fifo
File mapserver-5.4.2-r1.orig/.ipc_out is a fifo while file
mapserver-5.4.2-r1.edit/.ipc_out is a fifo
diff -ur mapserver-5.4.2-r1.orig/temp/environment
mapserver-5.4.2-r1.edit/temp/environment
--- mapserver-5.4.2-r1.orig/temp/environment 2010-12-23
21:25:57.216256714 -0600
+++ mapserver-5.4.2-r1.edit/temp/environment 2010-12-23
21:29:29.858254479 -0600

@@ -297,7 +297,7 @@
declare -x
S="/var/tmp/portage/sci-geosciences/mapserver-5.4.2-r1/work/mapserver-5.4.2"

declare -x SANDBOX_DEBUG="0"
declare -x SANDBOX_DENY=""
-declare -x SANDBOX_PID="2383"
+declare -x SANDBOX_PID="3207"
declare -x
SANDBOX_PREDICT="/var/tmp/portage/sci-geosciences/mapserver-5.4.2-r1/homedir:/dev/crypto:/var/cache/fontconfig"

declare -x SANDBOX_READ="/:/dev/stdin:/var/tmp"
declare -x SANDBOX_VERBOSE="1"
@@ -9371,7 +9371,7 @@
}
src_unpack ()
{
- default
+ ruby_src_unpack "$@"
}
strip-flags ()
{

gentoo-x86 commit in sci-geosciences/mapserver: mapserver-5.4.2-r1.ebuild ChangeLog
 

On 24/12/10 02:18, Arfrever Frehtes Taifersar Arahesis wrote:
> What do you mean about python.eclass?
> python.eclass doesn't define python_src_unpack().
>

No it doesn't, but calling the default() function in a phase will make
the default phase be called. And this is implemented in the python.eclass.

gentoo-x86 commit in sci-geosciences/mapserver: mapserver-5.4.2-r1.ebuild ChangeLog
 

On 12/23/2010 07:35 PM, Jeremy Olexa wrote:
> mapserver-5.4.2-r1.edit/temp/environment
> --- mapserver-5.4.2-r1.orig/temp/environment 2010-12-23
> 21:25:57.216256714 -0600
> +++ mapserver-5.4.2-r1.edit/temp/environment 2010-12-23
> 21:29:29.858254479 -0600
> @@ -297,7 +297,7 @@
> declare -x
> S="/var/tmp/portage/sci-geosciences/mapserver-5.4.2-r1/work/mapserver-5.4.2"
>
> declare -x SANDBOX_DEBUG="0"
> declare -x SANDBOX_DENY=""
> -declare -x SANDBOX_PID="2383"
> +declare -x SANDBOX_PID="3207"
> declare -x
> SANDBOX_PREDICT="/var/tmp/portage/sci-geosciences/mapserver-5.4.2-r1/homedir:/dev/crypto:/var/cache/fontconfig"
>
> declare -x SANDBOX_READ="/:/dev/stdin:/var/tmp"
> declare -x SANDBOX_VERBOSE="1"
> @@ -9371,7 +9371,7 @@
> }
> src_unpack ()
> {
> - default
> + ruby_src_unpack "$@"
> }
> strip-flags ()
> {

If you search for src_unpack inside that the environment file for
mapserver-5.4.2-r1.ebuild, then you'll find a grand total of 5
src_unpack implementations:

base_src_unpack
distutils_src_unpack
perl-module_src_unpack
php-ext-source-r1_src_unpack
ruby_src_unpack

I didn't check if all of them are actually exported, but at least
base.eclass seems to export src_unpack by default.
--
Thanks,
Zac

gentoo-x86 commit in sci-geosciences/mapserver: mapserver-5.4.2-r1.ebuild ChangeLog
 

On 24/12/10 20:46, Zac Medico wrote:
> On 12/23/2010 07:35 PM, Jeremy Olexa wrote:
>> mapserver-5.4.2-r1.edit/temp/environment
>> --- mapserver-5.4.2-r1.orig/temp/environment 2010-12-23
>> 21:25:57.216256714 -0600
>> +++ mapserver-5.4.2-r1.edit/temp/environment 2010-12-23
>> 21:29:29.858254479 -0600
>> @@ -297,7 +297,7 @@
>> declare -x
>> S="/var/tmp/portage/sci-geosciences/mapserver-5.4.2-r1/work/mapserver-5.4.2"
>>
>> declare -x SANDBOX_DEBUG="0"
>> declare -x SANDBOX_DENY=""
>> -declare -x SANDBOX_PID="2383"
>> +declare -x SANDBOX_PID="3207"
>> declare -x
>> SANDBOX_PREDICT="/var/tmp/portage/sci-geosciences/mapserver-5.4.2-r1/homedir:/dev/crypto:/var/cache/fontconfig"
>>
>> declare -x SANDBOX_READ="/:/dev/stdin:/var/tmp"
>> declare -x SANDBOX_VERBOSE="1"
>> @@ -9371,7 +9371,7 @@
>> }
>> src_unpack ()
>> {
>> - default
>> + ruby_src_unpack "$@"
>> }
>> strip-flags ()
>> {
>
> If you search for src_unpack inside that the environment file for
> mapserver-5.4.2-r1.ebuild, then you'll find a grand total of 5
> src_unpack implementations:
>
> base_src_unpack
> distutils_src_unpack
> perl-module_src_unpack
> php-ext-source-r1_src_unpack
> ruby_src_unpack
>
> I didn't check if all of them are actually exported, but at least
> base.eclass seems to export src_unpack by default.

You both are right.
I added it at first because the sources didn't got unpacked correctly,
so I assumed any of the eclass doing something I don't want. So I was
lazy and instead defining the phase correctly I justed used the
default() from python.eclass. I removed it again, because it is working
without. Probably I hit the issue during my try to move from ruby.eclass
to ruby-ng.eclass.

Thanks for noting all this, justin

gentoo-x86 commit in sci-geosciences/mapserver: mapserver-5.4.2-r1.ebuild ChangeLog
 

On 12/24/2010 11:19 AM, Justin (jlec) Lecher wrote:
> On 24/12/10 02:18, Arfrever Frehtes Taifersar Arahesis wrote:
>> What do you mean about python.eclass?
>> python.eclass doesn't define python_src_unpack().
>>
>
> No it doesn't, but calling the default() function in a phase will make
> the default phase be called. And this is implemented in the python.eclass.
>

default calls the PM implementation, not the eclass implementation.

From PMS:

default

Calls the default_ function for the current phase (see section 10.1.17).
Must not be called if the default_ function does not exist for the
current phase in the current EAPI. Only available in EAPIs listed in
table 12.14.

10.1.17:

DEFAULT-In EAPIs listed in table 10.8 as supporting default_ phase
functions, a function named default_(phase) that behaves as the default
implementation for that EAPI shall be defined when executing any ebuild
phase listed in the table. Ebuilds must not call these functions except
when in the phase in question.

Regards,
Petteri

gentoo-x86 commit in sci-geosciences/mapserver: mapserver-5.4.2-r1.ebuild ChangeLog
 

On 26/12/10 16:20, Petteri Räty wrote:
> On 12/24/2010 11:19 AM, Justin (jlec) Lecher wrote:
>> On 24/12/10 02:18, Arfrever Frehtes Taifersar Arahesis wrote:
>>> What do you mean about python.eclass?
>>> python.eclass doesn't define python_src_unpack().
>>>
>>
>> No it doesn't, but calling the default() function in a phase will make
>> the default phase be called. And this is implemented in the python.eclass.
>>
>
> default calls the PM implementation, not the eclass implementation.
>
> From PMS:
>
> default
>
> Calls the default_ function for the current phase (see section 10.1.17).
> Must not be called if the default_ function does not exist for the
> current phase in the current EAPI. Only available in EAPIs listed in
> table 12.14.
>
> 10.1.17:
>
> DEFAULT-In EAPIs listed in table 10.8 as supporting default_ phase
> functions, a function named default_(phase) that behaves as the default
> implementation for that EAPI shall be defined when executing any ebuild
> phase listed in the table. Ebuilds must not call these functions except
> when in the phase in question.
>
> Regards,
> Petteri
>

Thanks for clarification, Petteri.
I only knows this from the python dev guide, so I assumed it to come
from python.eclass.

justin