Re: [gentoo-dev] Minor changes in python.eclass and distutils.eclass
2010-07-06 01:37:01 Mike Frysinger napisał(a): On Monday, July 05, 2010 19:31:59 Arfrever Frehtes Taifersar Arahesis wrote: Only 84 ebuilds in the tree set NEED_PYTHON variable. NEED_PYTHON can be used only in EAPI =2. python_pkg_setup() is optional in EAPI =3. Exactly 0 ebuilds, which use NEED_PYTHON, also call python_pkg_setup(). EXPORT_FUNCTIONS wasn't ommitted during this calculation. well then, it shouldnt be hard for you to fix that small number and drop this crap from the tree. your global scope echo still doesnt require color support. everyone is telling you for obvious reasons to stop doing this. so do it already. NEED_PYTHON has been removed from all ebuilds in gentoo-x86, so now there shouldn't be any problems with printing this deprecation warning in global scope. -- Arfrever Frehtes Taifersar Arahesis signature.asc Description: This is a digitally signed message part.
Re: [gentoo-dev] Minor changes in python.eclass and distutils.eclass
On 07/06/2010 02:18 AM, Jeroen Roovers wrote: On Mon, 05 Jul 2010 22:50:56 + Jorge Manuel B. S. Vicetto jmbsvice...@gentoo.org wrote: I'm not going to delve into the details that have been addressed all other this thread. Instead I'll just address one small issue. The use of *minor* in the title of this thread and the sheer size of the patch attached are not compatible. Please don't label changes such as these as *minor* in future instances. Another major flaw on its own is the sheer size of the patch (it hadn't been explicitly mentioned to my knowledge). Aren't we supposed to commit early and often in open source projects instead of dumping all your changes on reviewers at once (first fork and later merge)? And if you're still going to do it, a the huge patch should only be necessary to establish a single sweeping change, not many small changes. Indeed the patch should be splitted if you expect people to reasonably review it. Petteri signature.asc Description: OpenPGP digital signature
[gentoo-dev] Minor changes in python.eclass and distutils.eclass
These minor changes in python.eclass and distutils.eclass have been already
reviewed on alias of Gentoo Python Project. It's recommended to be familiar
with internals of current code before trying to understand these minor changes.
Suggestions about indentation and quoting will be rejected.
--
Arfrever Frehtes Taifersar Arahesis
--- python.eclass
+++ python.eclass
@@ -161,9 +161,26 @@
_parse_PYTHON_DEPEND
elif [[ -n ${NEED_PYTHON} ]]; then
if ! has ${EAPI:-0} 0 1 2; then
- eerror Use PYTHON_DEPEND instead of NEED_PYTHON.
- die NEED_PYTHON cannot be used in this EAPI
+ eerror Use PYTHON_DEPEND variable instead of NEED_PYTHON variable.
+ die NEED_PYTHON variable cannot be used in this EAPI
fi
+
+ if [[ ${NOCOLOR:-false} =~ ^(false|no)$ ]]; then
+ _RED=$'\e[1;31m'
+ _NORMAL=$'\e[0m'
+ else
+ _RED=
+ _NORMAL=
+ fi
+
+ echo
+ echo ${_RED}*${_NORMAL} ${_RED}Deprecation Warning: NEED_PYTHON variable is deprecated and will be banned on 2010-10-01.${_NORMAL}
+ echo ${_RED}*${_NORMAL} ${_RED}Use PYTHON_DEPEND variable instead of NEED_PYTHON variable.${_NORMAL}
+ echo ${_RED}*${_NORMAL} ${_RED}The ebuild needs to be fixed. Please report a bug, if it has not been already reported.${_NORMAL}
+ echo
+
+ unset _BOLD _NORMAL
+
_PYTHON_ATOMS=(=dev-lang/python-${NEED_PYTHON})
DEPEND+=${DEPEND:+ }${_python_ato...@]}
RDEPEND+=${RDEPEND:+ }${_python_ato...@]}
@@ -243,6 +260,10 @@
fi
}
+_python_abi-specific_local_scope() {
+ [[ ${funcna...@]:2} =~ (_python_final_sanity_checks|python_execute_function|python_mod_optimize|python_mod_cleanup) ]]
+}
+
_python_initialize_prefix_variables() {
if has ${EAPI:-0} 0 1 2; then
if [[ -n ${ROOT} -z ${EROOT} ]]; then
@@ -254,10 +275,10 @@
fi
}
-unset PYTHON_SANITY_CHECKS
+unset PYTHON_SANITY_CHECKS_EXECUTED PYTHON_SKIP_SANITY_CHECKS
_python_initial_sanity_checks() {
- if [[ $(declare -p PYTHON_SANITY_CHECKS 2 /dev/null) != declare -- PYTHON_SANITY_CHECKS=* ]]; then
+ if [[ $(declare -p PYTHON_SANITY_CHECKS_EXECUTED 2 /dev/null) != declare -- PYTHON_SANITY_CHECKS_EXECUTED=* || ${funcna...@]:1} =~ (python_set_active_version|python_pkg_setup) -z ${PYTHON_SKIP_SANITY_CHECKS} ]]; then
# Ensure that /usr/bin/python and /usr/bin/python-config are valid.
if [[ $(readlink ${EPREFIX}/usr/bin/python) != python-wrapper ]]; then
eerror '${EPREFIX}/usr/bin/python' is not valid symlink.
@@ -273,7 +294,7 @@
}
_python_final_sanity_checks() {
- if ! _python_implementation [[ $(declare -p PYTHON_SANITY_CHECKS 2 /dev/null) != declare -- PYTHON_SANITY_CHECKS=* ]]; then
+ if ! _python_implementation [[ $(declare -p PYTHON_SANITY_CHECKS_EXECUTED 2 /dev/null) != declare -- PYTHON_SANITY_CHECKS_EXECUTED=* || ${funcna...@]:1} =~ (python_set_active_version|python_pkg_setup) -z ${PYTHON_SKIP_SANITY_CHECKS} ]]; then
local PYTHON_ABI=${PYTHON_ABI}
for PYTHON_ABI in ${PYTHON_ABIS-${PYTHON_ABI}}; do
# Ensure that appropriate version of Python is installed.
@@ -293,7 +314,7 @@
fi
done
fi
- PYTHON_SANITY_CHECKS=1
+ PYTHON_SANITY_CHECKS_EXECUTED=1
}
_python_set_color_variables() {
@@ -316,6 +337,14 @@
unset PYTHON_PKG_SETUP_EXECUTED
+_python_check_python_pkg_setup_execution() {
+ [[ ${funcna...@]:1} =~ (python_set_active_version|python_pkg_setup) ]] return
+
+ if ! has ${EAPI:-0} 0 1 2 3 [[ -z ${PYTHON_PKG_SETUP_EXECUTED} ]]; then
+ die python_pkg_setup() not called
+ fi
+}
+
# @FUNCTION: python_pkg_setup
# @DESCRIPTION:
# Perform sanity checks and initialize environment.
@@ -328,6 +357,10 @@
# Check if phase is pkg_setup().
[[ ${EBUILD_PHASE} != setup ]] die ${FUNCNAME}() can be used only in pkg_setup() phase
+ if [[ $# -ne 0 ]]; then
+ die ${FUNCNAME}() does not accept arguments
+ fi
+
if _python_package_supporting_installation_for_multiple_python_abis; then
_python_calculate_PYTHON_ABIS
export EPYTHON=$(PYTHON -f)
@@ -368,7 +401,7 @@
}
if _python_package_supporting_installation_for_multiple_python_abis; then
- python_execute_function -q python_pkg_setup_check_USE_flags
+ PYTHON_SKIP_SANITY_CHECKS=1 python_execute_function -q python_pkg_setup_check_USE_flags
else
python_pkg_setup_check_USE_flags
fi
@@ -388,6 +421,8 @@
# @DESCRIPTION:
# Convert shebangs in specified files. Directories can be specified only with --recursive option.
python_convert_shebangs() {
+ _python_check_python_pkg_setup_execution
+
local argument file files=() only_executables=0 python_version quiet=0 recursive=0
while (($#)); do
@@ -466,6 +501,7 @@
# @DESCRIPTION:
# Delete needless files in installation image.
python_clean_installation_image() {
+ _python_check_python_pkg_setup_execution
_python_initialize_prefix_variables
local file files=() quiet=0
@@ -538,15 +574,23 @@
# src_prepare, src_configure, src_compile, src_test, src_install.
if ! has ${EAPI:-0} 0 1; then
python_src_prepare() {
+ _python_check_python_pkg_setup_execution
+
if !
Re: [gentoo-dev] Minor changes in python.eclass and distutils.eclass
On 07/05/2010 06:23 PM, Arfrever Frehtes Taifersar Arahesis wrote: These minor changes in python.eclass and distutils.eclass have been already reviewed on alias of Gentoo Python Project. It's recommended to be familiar with internals of current code before trying to understand these minor changes. Suggestions about indentation and quoting will be rejected. You have been already told to get rid of all the color customizations in the python eclasses here: http://bugs.gentoo.org/show_bug.cgi?id=309057#c2 http://bugs.gentoo.org/show_bug.cgi?id=309057#c3 [ .. ] http://bugs.gentoo.org/show_bug.cgi?id=309057#c5 The bug was wrongly closed as fixed, as it's not really fixed before it's all punted - Samuli
Re: [gentoo-dev] Minor changes in python.eclass and distutils.eclass
2010-07-05 18:13:26 Samuli Suominen napisał(a): On 07/05/2010 06:23 PM, Arfrever Frehtes Taifersar Arahesis wrote: These minor changes in python.eclass and distutils.eclass have been already reviewed on alias of Gentoo Python Project. It's recommended to be familiar with internals of current code before trying to understand these minor changes. Suggestions about indentation and quoting will be rejected. You have been already told to get rid of all the color customizations in the python eclasses here: http://bugs.gentoo.org/show_bug.cgi?id=309057#c2 http://bugs.gentoo.org/show_bug.cgi?id=309057#c3 [ .. ] http://bugs.gentoo.org/show_bug.cgi?id=309057#c5 The bug was wrongly closed as fixed, as it's not really fixed before it's all punted Colors can be used with echo. -- Arfrever Frehtes Taifersar Arahesis signature.asc Description: This is a digitally signed message part.
Re: [gentoo-dev] Minor changes in python.eclass and distutils.eclass
On 07/05/2010 07:17 PM, Arfrever Frehtes Taifersar Arahesis wrote: 2010-07-05 18:13:26 Samuli Suominen napisał(a): On 07/05/2010 06:23 PM, Arfrever Frehtes Taifersar Arahesis wrote: These minor changes in python.eclass and distutils.eclass have been already reviewed on alias of Gentoo Python Project. It's recommended to be familiar with internals of current code before trying to understand these minor changes. Suggestions about indentation and quoting will be rejected. You have been already told to get rid of all the color customizations in the python eclasses here: http://bugs.gentoo.org/show_bug.cgi?id=309057#c2 http://bugs.gentoo.org/show_bug.cgi?id=309057#c3 [ .. ] http://bugs.gentoo.org/show_bug.cgi?id=309057#c5 The bug was wrongly closed as fixed, as it's not really fixed before it's all punted Colors can be used with echo. Stop using echo for output and switch to standard output functions, like einfo/eerror/elog/... like told in http://bugs.gentoo.org/show_bug.cgi?id=309057#c5
Re: [gentoo-dev] Minor changes in python.eclass and distutils.eclass
2010-07-05 18:26:40 Samuli Suominen napisał(a): On 07/05/2010 07:17 PM, Arfrever Frehtes Taifersar Arahesis wrote: 2010-07-05 18:13:26 Samuli Suominen napisał(a): On 07/05/2010 06:23 PM, Arfrever Frehtes Taifersar Arahesis wrote: These minor changes in python.eclass and distutils.eclass have been already reviewed on alias of Gentoo Python Project. It's recommended to be familiar with internals of current code before trying to understand these minor changes. Suggestions about indentation and quoting will be rejected. You have been already told to get rid of all the color customizations in the python eclasses here: http://bugs.gentoo.org/show_bug.cgi?id=309057#c2 http://bugs.gentoo.org/show_bug.cgi?id=309057#c3 [ .. ] http://bugs.gentoo.org/show_bug.cgi?id=309057#c5 The bug was wrongly closed as fixed, as it's not really fixed before it's all punted Colors can be used with echo. Stop using echo for output and switch to standard output functions, like einfo/eerror/elog/... like told in http://bugs.gentoo.org/show_bug.cgi?id=309057#c5 You should read relevant part of comment #7: The colors can of course be continued to be used in outputs that are purely build time outputting and not for communicating things for users like what cmake builds do. python.eclass uses colors for build time outputting, which doesn't communicate anything for users. -- Arfrever Frehtes Taifersar Arahesis signature.asc Description: This is a digitally signed message part.
Re: [gentoo-dev] Minor changes in python.eclass and distutils.eclass
On Mon, 5 Jul 2010 18:34:38 +0200 Arfrever Frehtes Taifersar Arahesis arfre...@gentoo.org wrote: You should read relevant part of comment #7: The colors can of course be continued to be used in outputs that are purely build time outputting and not for communicating things for users like what cmake builds do. python.eclass uses colors for build time outputting, which doesn't communicate anything for users. Well stop it. You're cluttering up your eclass with pointless junk under the mistaken impression that more complicated eclasses are better. If you have a desperate need for a coloured output that isn't doable using the existing e* functions, make your case for either a new EAPI or for a special, dedicated eclass. Besides, if you're not communicating anything for users, why have it there? -- Ciaran McCreesh signature.asc Description: PGP signature
Re: [gentoo-dev] Minor changes in python.eclass and distutils.eclass
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Dne 5.7.2010 18:34, Arfrever Frehtes Taifersar Arahesis napsal(a):
2010-07-05 18:26:40 Samuli Suominen napisał(a):
On 07/05/2010 07:17 PM, Arfrever Frehtes Taifersar Arahesis wrote:
2010-07-05 18:13:26 Samuli Suominen napisał(a):
On 07/05/2010 06:23 PM, Arfrever Frehtes Taifersar Arahesis wrote:
These minor changes in python.eclass and distutils.eclass have been
already
reviewed on alias of Gentoo Python Project. It's recommended to be
familiar
with internals of current code before trying to understand these minor
changes.
Suggestions about indentation and quoting will be rejected.
You have been already told to get rid of all the color customizations in
the python eclasses here:
http://bugs.gentoo.org/show_bug.cgi?id=309057#c2
http://bugs.gentoo.org/show_bug.cgi?id=309057#c3
[ .. ]
http://bugs.gentoo.org/show_bug.cgi?id=309057#c5
The bug was wrongly closed as fixed, as it's not really fixed before
it's all punted
Colors can be used with echo.
Stop using echo for output and switch to standard output functions, like
einfo/eerror/elog/... like told in
http://bugs.gentoo.org/show_bug.cgi?id=309057#c5
You should read relevant part of comment #7:
The colors can of course be continued to be used in outputs that are purely
build
time outputting and not for communicating things for users like what cmake
builds do.
python.eclass uses colors for build time outputting, which doesn't
communicate anything for users.
+ echo ${_RED}*${_NORMAL} ${_RED}Deprecation Warning: NEED_PYTHON
variable is deprecated and will be banned on 2010-10-01.${_NORMAL}
+ echo ${_RED}*${_NORMAL} ${_RED}Use PYTHON_DEPEND variable instead of
NEED_PYTHON variable.${_NORMAL}
+ echo ${_RED}*${_NORMAL} ${_RED}The ebuild needs to be fixed. Please
report a bug, if it has not been already reported.${_NORMAL}
The above is build outputting since when?
-BEGIN PGP SIGNATURE-
Version: GnuPG v2.0.15 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
iEYEARECAAYFAkwyCfkACgkQHB6c3gNBRYd2yQCgrpR4HKMnCIqyRbN9eAomZwRK
b8oAnjBW16mkxb7B5hpbqZXN5IQ4W4oy
=xV10
-END PGP SIGNATURE-
Re: [gentoo-dev] Minor changes in python.eclass and distutils.eclass
2010-07-05 18:36:09 Tomáš Chvátal napisał(a):
Dne 5.7.2010 18:34, Arfrever Frehtes Taifersar Arahesis napsal(a):
2010-07-05 18:26:40 Samuli Suominen napisał(a):
On 07/05/2010 07:17 PM, Arfrever Frehtes Taifersar Arahesis wrote:
2010-07-05 18:13:26 Samuli Suominen napisał(a):
On 07/05/2010 06:23 PM, Arfrever Frehtes Taifersar Arahesis wrote:
These minor changes in python.eclass and distutils.eclass have been
already
reviewed on alias of Gentoo Python Project. It's recommended to be
familiar
with internals of current code before trying to understand these minor
changes.
Suggestions about indentation and quoting will be rejected.
You have been already told to get rid of all the color customizations in
the python eclasses here:
http://bugs.gentoo.org/show_bug.cgi?id=309057#c2
http://bugs.gentoo.org/show_bug.cgi?id=309057#c3
[ .. ]
http://bugs.gentoo.org/show_bug.cgi?id=309057#c5
The bug was wrongly closed as fixed, as it's not really fixed before
it's all punted
Colors can be used with echo.
Stop using echo for output and switch to standard output functions, like
einfo/eerror/elog/... like told in
http://bugs.gentoo.org/show_bug.cgi?id=309057#c5
You should read relevant part of comment #7:
The colors can of course be continued to be used in outputs that are
purely build
time outputting and not for communicating things for users like what cmake
builds do.
python.eclass uses colors for build time outputting, which doesn't
communicate anything for users.
+ echo ${_RED}*${_NORMAL} ${_RED}Deprecation Warning: NEED_PYTHON
variable is deprecated and will be banned on 2010-10-01.${_NORMAL}
+ echo ${_RED}*${_NORMAL} ${_RED}Use PYTHON_DEPEND variable instead of
NEED_PYTHON variable.${_NORMAL}
+ echo ${_RED}*${_NORMAL} ${_RED}The ebuild needs to be fixed. Please
report a bug, if it has not been already reported.${_NORMAL}
The above is build outputting since when?
The colored, non-logged output in deprecation warnings is used as exception to
increase
the chance that ebuild maintainers will be notified earlier about the necessity
of changes
in given ebuilds.
--
Arfrever Frehtes Taifersar Arahesis
signature.asc
Description: This is a digitally signed message part.
Re: [gentoo-dev] Minor changes in python.eclass and distutils.eclass
2010-07-05 18:37:37 Ciaran McCreesh napisał(a): You're cluttering up your eclass with pointless junk under the mistaken impression that more complicated eclasses are better. These eclasses don't contain any unnecessary code (except deprecated functions/variables). Besides, if you're not communicating anything for users, why have it there? The colored output is intended and useful for ebuild maintainers. -- Arfrever Frehtes Taifersar Arahesis signature.asc Description: This is a digitally signed message part.
Re: [gentoo-dev] Minor changes in python.eclass and distutils.eclass
On Mon, Jul 05, 2010 at 07:01:27PM +0200, Arfrever Frehtes Taifersar Arahesis
wrote:
2010-07-05 18:36:09 Tomáš Chvátal napisał(a):
Dne 5.7.2010 18:34, Arfrever Frehtes Taifersar Arahesis napsal(a):
python.eclass uses colors for build time outputting, which doesn't
communicate anything for users.
+ echo ${_RED}*${_NORMAL} ${_RED}Deprecation Warning: NEED_PYTHON
variable is deprecated and will be banned on 2010-10-01.${_NORMAL}
+ echo ${_RED}*${_NORMAL} ${_RED}Use PYTHON_DEPEND variable instead of
NEED_PYTHON variable.${_NORMAL}
+ echo ${_RED}*${_NORMAL} ${_RED}The ebuild needs to be fixed. Please
report a bug, if it has not been already reported.${_NORMAL}
The above is build outputting since when?
The colored, non-logged output in deprecation warnings is used as exception
to increase
the chance that ebuild maintainers will be notified earlier about the
necessity of changes
in given ebuilds.
einfo/ewarn/eerror output is repeated by default when emerge exits. By
not using einfo/ewarn/eerror, you are making it less likely that others
will be reading your deprecation notices.
Re: [gentoo-dev] Minor changes in python.eclass and distutils.eclass
On Mon, Jul 05, 2010 at 07:38:32PM +0200, Harald van Dijk wrote:
On Mon, Jul 05, 2010 at 07:01:27PM +0200, Arfrever Frehtes Taifersar Arahesis
wrote:
2010-07-05 18:36:09 Tomáš Chvátal napisał(a):
Dne 5.7.2010 18:34, Arfrever Frehtes Taifersar Arahesis napsal(a):
python.eclass uses colors for build time outputting, which doesn't
communicate anything for users.
+ echo ${_RED}*${_NORMAL} ${_RED}Deprecation Warning: NEED_PYTHON
variable is deprecated and will be banned on 2010-10-01.${_NORMAL}
+ echo ${_RED}*${_NORMAL} ${_RED}Use PYTHON_DEPEND variable instead of
NEED_PYTHON variable.${_NORMAL}
+ echo ${_RED}*${_NORMAL} ${_RED}The ebuild needs to be fixed. Please
report a bug, if it has not been already reported.${_NORMAL}
The above is build outputting since when?
The colored, non-logged output in deprecation warnings is used as exception
to increase
the chance that ebuild maintainers will be notified earlier about the
necessity of changes
in given ebuilds.
einfo/ewarn/eerror output is repeated by default when emerge exits. By
not using einfo/ewarn/eerror, you are making it less likely that others
will be reading your deprecation notices.
Ugh. I see that you're using einfo already and suppressing its output.
In that case, my objection doesn't apply, but it's still nasty.
Re: [gentoo-dev] Minor changes in python.eclass and distutils.eclass
El lun, 05-07-2010 a las 19:38 +0200, Harald van Dijk escribió:
On Mon, Jul 05, 2010 at 07:01:27PM +0200, Arfrever Frehtes Taifersar Arahesis
wrote:
2010-07-05 18:36:09 Tomáš Chvátal napisał(a):
Dne 5.7.2010 18:34, Arfrever Frehtes Taifersar Arahesis napsal(a):
python.eclass uses colors for build time outputting, which doesn't
communicate anything for users.
+ echo ${_RED}*${_NORMAL} ${_RED}Deprecation Warning: NEED_PYTHON
variable is deprecated and will be banned on 2010-10-01.${_NORMAL}
+ echo ${_RED}*${_NORMAL} ${_RED}Use PYTHON_DEPEND variable instead of
NEED_PYTHON variable.${_NORMAL}
+ echo ${_RED}*${_NORMAL} ${_RED}The ebuild needs to be fixed. Please
report a bug, if it has not been already reported.${_NORMAL}
The above is build outputting since when?
The colored, non-logged output in deprecation warnings is used as exception
to increase
the chance that ebuild maintainers will be notified earlier about the
necessity of changes
in given ebuilds.
einfo/ewarn/eerror output is repeated by default when emerge exits. By
not using einfo/ewarn/eerror, you are making it less likely that others
will be reading your deprecation notices.
I fully agree with this, I think that using ewarn would be much more
useful for all of us
signature.asc
Description: This is a digitally signed message part
Re: [gentoo-dev] Minor changes in python.eclass and distutils.eclass
Arfrever Frehtes Taifersar Arahesis arfre...@gentoo.org said: 2010-07-05 18:26:40 Samuli Suominen napisał(a): On 07/05/2010 07:17 PM, Arfrever Frehtes Taifersar Arahesis wrote: 2010-07-05 18:13:26 Samuli Suominen napisał(a): On 07/05/2010 06:23 PM, Arfrever Frehtes Taifersar Arahesis wrote: These minor changes in python.eclass and distutils.eclass have been already reviewed on alias of Gentoo Python Project. It's recommended to be familiar with internals of current code before trying to understand these minor changes. Suggestions about indentation and quoting will be rejected. You have been already told to get rid of all the color customizations in the python eclasses here: http://bugs.gentoo.org/show_bug.cgi?id=309057#c2 http://bugs.gentoo.org/show_bug.cgi?id=309057#c3 [ .. ] http://bugs.gentoo.org/show_bug.cgi?id=309057#c5 The bug was wrongly closed as fixed, as it's not really fixed before it's all punted Colors can be used with echo. Stop using echo for output and switch to standard output functions, like einfo/eerror/elog/... like told in http://bugs.gentoo.org/show_bug.cgi?id=309057#c5 You should read relevant part of comment #7: The colors can of course be continued to be used in outputs that are purely build time outputting and not for communicating things for users like what cmake builds do. python.eclass uses colors for build time outputting, which doesn't communicate anything for users. Everyone else has already made valid points. I'm just picking this one to reply to now. Please remove the colors you have added. If you need a new function, say eqawarn, we should have that added in the next EAPI with a description of when and where to use it. In the meantime, Petteri proposed a nice solution awhile back that would centralize this so it is not a one-off hack. Here is a link to his original proposal: http://archives.gentoo.org/gentoo-dev/msg_44d395a1b887468051a1e1c049e99ba3.xml Thanks, -- Mark Loeser email - halcy0n AT gentoo DOT org email - mark AT halcy0n DOT com web - http://www.halcy0n.com signature.asc Description: Digital signature
Re: [gentoo-dev] Minor changes in python.eclass and distutils.eclass
On 07/05/2010 08:55 PM, Pacho Ramos wrote:
El lun, 05-07-2010 a las 19:38 +0200, Harald van Dijk escribió:
On Mon, Jul 05, 2010 at 07:01:27PM +0200, Arfrever Frehtes Taifersar
Arahesis wrote:
2010-07-05 18:36:09 Tomáš Chvátal napisał(a):
Dne 5.7.2010 18:34, Arfrever Frehtes Taifersar Arahesis napsal(a):
python.eclass uses colors for build time outputting, which doesn't
communicate anything for users.
+ echo ${_RED}*${_NORMAL} ${_RED}Deprecation Warning: NEED_PYTHON
variable is deprecated and will be banned on 2010-10-01.${_NORMAL}
+ echo ${_RED}*${_NORMAL} ${_RED}Use PYTHON_DEPEND variable instead of
NEED_PYTHON variable.${_NORMAL}
+ echo ${_RED}*${_NORMAL} ${_RED}The ebuild needs to be fixed. Please
report a bug, if it has not been already reported.${_NORMAL}
The above is build outputting since when?
The colored, non-logged output in deprecation warnings is used as exception
to increase
the chance that ebuild maintainers will be notified earlier about the
necessity of changes
in given ebuilds.
einfo/ewarn/eerror output is repeated by default when emerge exits. By
not using einfo/ewarn/eerror, you are making it less likely that others
will be reading your deprecation notices.
I fully agree with this, I think that using ewarn would be much more
useful for all of us
We should not be spamming our users with deprecation notices. ewarn is
for messages that users should take action for. People can for example
have configured to receive email on these messages.
Regards,
Petteri
signature.asc
Description: OpenPGP digital signature
Re: [gentoo-dev] Minor changes in python.eclass and distutils.eclass
On 07/05/2010 06:23 PM, Arfrever Frehtes Taifersar Arahesis wrote: These minor changes in python.eclass and distutils.eclass have been already reviewed on alias of Gentoo Python Project. It's recommended to be familiar with internals of current code before trying to understand these minor changes. If you are looking for review on gentoo-dev you should include something on what the patch is supposed to do. Suggestions about indentation and quoting will be rejected. Do you think our indentation rules don't concern you? In general the email reads like you are not really looking for feedback on the patch so why bother gentoo-dev at all? Regards, Petteri signature.asc Description: OpenPGP digital signature
Re: [gentoo-dev] Minor changes in python.eclass and distutils.eclass
2010-07-05 20:00:11 Mark Loeser napisał(a): Arfrever Frehtes Taifersar Arahesis arfre...@gentoo.org said: 2010-07-05 18:26:40 Samuli Suominen napisał(a): On 07/05/2010 07:17 PM, Arfrever Frehtes Taifersar Arahesis wrote: 2010-07-05 18:13:26 Samuli Suominen napisał(a): On 07/05/2010 06:23 PM, Arfrever Frehtes Taifersar Arahesis wrote: These minor changes in python.eclass and distutils.eclass have been already reviewed on alias of Gentoo Python Project. It's recommended to be familiar with internals of current code before trying to understand these minor changes. Suggestions about indentation and quoting will be rejected. You have been already told to get rid of all the color customizations in the python eclasses here: http://bugs.gentoo.org/show_bug.cgi?id=309057#c2 http://bugs.gentoo.org/show_bug.cgi?id=309057#c3 [ .. ] http://bugs.gentoo.org/show_bug.cgi?id=309057#c5 The bug was wrongly closed as fixed, as it's not really fixed before it's all punted Colors can be used with echo. Stop using echo for output and switch to standard output functions, like einfo/eerror/elog/... like told in http://bugs.gentoo.org/show_bug.cgi?id=309057#c5 You should read relevant part of comment #7: The colors can of course be continued to be used in outputs that are purely build time outputting and not for communicating things for users like what cmake builds do. python.eclass uses colors for build time outputting, which doesn't communicate anything for users. Everyone else has already made valid points. I'm just picking this one to reply to now. Please remove the colors you have added. If you need a new function, say eqawarn, we should have that added in the next EAPI with a description of when and where to use it. In case of the colored message added in this patch, if einfo/elog/ewarn/eqawarn/eerror was used, then its output wouldn't be logged by Portage. -- Arfrever Frehtes Taifersar Arahesis signature.asc Description: This is a digitally signed message part.
Re: [gentoo-dev] Minor changes in python.eclass and distutils.eclass
Arfrever Frehtes Taifersar Arahesis arfre...@gentoo.org said: 2010-07-05 20:00:11 Mark Loeser napisał(a): Everyone else has already made valid points. I'm just picking this one to reply to now. Please remove the colors you have added. If you need a new function, say eqawarn, we should have that added in the next EAPI with a description of when and where to use it. In case of the colored message added in this patch, if einfo/elog/ewarn/eqawarn/eerror was used, then its output wouldn't be logged by Portage. I don't understand what you are trying to say. The QA team has decided that the coloring should be removed from the python eclass and a centralized generic solution should be proposed and agreed upon. Thanks, -- Mark Loeser email - halcy0n AT gentoo DOT org email - mark AT halcy0n DOT com web - http://www.halcy0n.com signature.asc Description: Digital signature
Re: [gentoo-dev] Minor changes in python.eclass and distutils.eclass
On Mon, Jul 05, 2010 at 05:23:58PM +0200, Arfrever Frehtes Taifersar Arahesis wrote: These minor changes in python.eclass and distutils.eclass have been already reviewed on alias of Gentoo Python Project. It's recommended to be familiar with internals of current code before trying to understand these minor changes. Suggestions about indentation and quoting will be rejected. I don't understand. eclass patches are supposed to be sent to -dev as RFCs, not as hey, I did this, but if you want to comment, don't, I don't care. You should welcome critisism, your fellow devs want whats best for Gentoo, they are not after you :) Truth is, all that weird coloring is messing up the eclass. At the very least you could have defined your own epinfo function or something to cover it up. Or pushed for Peterri's eqawarn solution. -- Arfrever Frehtes Taifersar Arahesis -- Alex Alexander :: wired Gentoo Developer www.linuxized.com pgpnlC6WnX40e.pgp Description: PGP signature
Re: [gentoo-dev] Minor changes in python.eclass and distutils.eclass
2010-07-05 20:26:19 Petteri Räty napisał(a): On 07/05/2010 06:23 PM, Arfrever Frehtes Taifersar Arahesis wrote: These minor changes in python.eclass and distutils.eclass have been already reviewed on alias of Gentoo Python Project. It's recommended to be familiar with internals of current code before trying to understand these minor changes. If you are looking for review on gentoo-dev you should include something on what the patch is supposed to do. In python.eclass: - Deprecation of NEED_PYTHON variable - Deletion of colors from 3 deprecated functions and some cleanup in these functions - Improvements in error messages, sanity checks and handling of arguments - First part of support for API of python.eclass in EAPI=4. - Some new, optional features in some functions (e.g. --base-path option in some python_get_*() functions) - Support for multiple paths in argument of --PYTHONPATH option of test helper functions In distutils.eclass: - Support for custom setup files specified in DISTUTILS_SETUP_FILES array - Deletion of colors from 2 deprecated functions and some cleanup in thse functions Suggestions about indentation and quoting will be rejected. Do you think our indentation rules don't concern you? I don't use indentation/quoting, which would violate any rules, but some people might try to enforce e.g. EAPI=3 instead of EAPI=3. -- Arfrever Frehtes Taifersar Arahesis signature.asc Description: This is a digitally signed message part.
Re: [gentoo-dev] Minor changes in python.eclass and distutils.eclass
On 07/05/2010 10:59 PM, Arfrever Frehtes Taifersar Arahesis wrote: I don't use indentation/quoting, which would violate any rules, but some people might try to enforce e.g. EAPI=3 instead of EAPI=3. You are implying you would never screw up accidentally. Regards, Petteri signature.asc Description: OpenPGP digital signature
Re: [gentoo-dev] Minor changes in python.eclass and distutils.eclass
2010-07-05 21:18:57 Mark Loeser napisał(a): Arfrever Frehtes Taifersar Arahesis arfre...@gentoo.org said: 2010-07-05 20:00:11 Mark Loeser napisał(a): Everyone else has already made valid points. I'm just picking this one to reply to now. Please remove the colors you have added. If you need a new function, say eqawarn, we should have that added in the next EAPI with a description of when and where to use it. In case of the colored message added in this patch, if einfo/elog/ewarn/eqawarn/eerror was used, then its output wouldn't be logged by Portage. I don't understand what you are trying to say. Portage doesn't log output of einfo/elog/ewarn/eqawarn/eerror called in global scope. -- Arfrever Frehtes Taifersar Arahesis signature.asc Description: This is a digitally signed message part.
Re: [gentoo-dev] Minor changes in python.eclass and distutils.eclass
On 07/05/2010 11:23 PM, Arfrever Frehtes Taifersar Arahesis wrote: 2010-07-05 21:18:57 Mark Loeser napisał(a): Arfrever Frehtes Taifersar Arahesis arfre...@gentoo.org said: 2010-07-05 20:00:11 Mark Loeser napisał(a): Everyone else has already made valid points. I'm just picking this one to reply to now. Please remove the colors you have added. If you need a new function, say eqawarn, we should have that added in the next EAPI with a description of when and where to use it. In case of the colored message added in this patch, if einfo/elog/ewarn/eqawarn/eerror was used, then its output wouldn't be logged by Portage. I don't understand what you are trying to say. Portage doesn't log output of einfo/elog/ewarn/eqawarn/eerror called in global scope. Can you point to the location of python.eclass / distutils.eclass where you need to send output from global scope, please?
Re: [gentoo-dev] Minor changes in python.eclass and distutils.eclass
2010-07-05 23:28:01 Samuli Suominen napisał(a): On 07/05/2010 11:23 PM, Arfrever Frehtes Taifersar Arahesis wrote: 2010-07-05 21:18:57 Mark Loeser napisał(a): Arfrever Frehtes Taifersar Arahesis arfre...@gentoo.org said: 2010-07-05 20:00:11 Mark Loeser napisał(a): Everyone else has already made valid points. I'm just picking this one to reply to now. Please remove the colors you have added. If you need a new function, say eqawarn, we should have that added in the next EAPI with a description of when and where to use it. In case of the colored message added in this patch, if einfo/elog/ewarn/eqawarn/eerror was used, then its output wouldn't be logged by Portage. I don't understand what you are trying to say. Portage doesn't log output of einfo/elog/ewarn/eqawarn/eerror called in global scope. Can you point to the location of python.eclass / distutils.eclass where you need to send output from global scope, please? It's at the beginning of this patch. NEED_PYTHON is parsed in global scope. -- Arfrever Frehtes Taifersar Arahesis signature.asc Description: This is a digitally signed message part.
Re: [gentoo-dev] Minor changes in python.eclass and distutils.eclass
On 07/06/2010 12:38 AM, Arfrever Frehtes Taifersar Arahesis wrote: 2010-07-05 23:28:01 Samuli Suominen napisał(a): On 07/05/2010 11:23 PM, Arfrever Frehtes Taifersar Arahesis wrote: 2010-07-05 21:18:57 Mark Loeser napisał(a): Arfrever Frehtes Taifersar Arahesis arfre...@gentoo.org said: 2010-07-05 20:00:11 Mark Loeser napisał(a): Everyone else has already made valid points. I'm just picking this one to reply to now. Please remove the colors you have added. If you need a new function, say eqawarn, we should have that added in the next EAPI with a description of when and where to use it. In case of the colored message added in this patch, if einfo/elog/ewarn/eqawarn/eerror was used, then its output wouldn't be logged by Portage. I don't understand what you are trying to say. Portage doesn't log output of einfo/elog/ewarn/eqawarn/eerror called in global scope. Can you point to the location of python.eclass / distutils.eclass where you need to send output from global scope, please? It's at the beginning of this patch. NEED_PYTHON is parsed in global scope. OK, so let me get this right... are trying to justify using all of the echo+custom colorization in the eclass by this one occurance, in basis of it doesn't get logged otherwise? Seems a bit far fetch. Also I'm not convinced if it's beneficial to log this.
Re: [gentoo-dev] Minor changes in python.eclass and distutils.eclass
2010-07-05 23:49:58 Samuli Suominen napisał(a): On 07/06/2010 12:38 AM, Arfrever Frehtes Taifersar Arahesis wrote: 2010-07-05 23:28:01 Samuli Suominen napisał(a): On 07/05/2010 11:23 PM, Arfrever Frehtes Taifersar Arahesis wrote: 2010-07-05 21:18:57 Mark Loeser napisał(a): Arfrever Frehtes Taifersar Arahesis arfre...@gentoo.org said: 2010-07-05 20:00:11 Mark Loeser napisał(a): Everyone else has already made valid points. I'm just picking this one to reply to now. Please remove the colors you have added. If you need a new function, say eqawarn, we should have that added in the next EAPI with a description of when and where to use it. In case of the colored message added in this patch, if einfo/elog/ewarn/eqawarn/eerror was used, then its output wouldn't be logged by Portage. I don't understand what you are trying to say. Portage doesn't log output of einfo/elog/ewarn/eqawarn/eerror called in global scope. Can you point to the location of python.eclass / distutils.eclass where you need to send output from global scope, please? It's at the beginning of this patch. NEED_PYTHON is parsed in global scope. OK, so let me get this right... are trying to justify using all of the echo+custom colorization in the eclass by this one occurance, in basis of it doesn't get logged otherwise? I was justifying using colors in this case (NEED_PYTHON). I'm completely against not using colors in status messages (e.g. Building of dev-python/setuptools-0.6.13 with CPython 3.1...), which are more useful, when they are easily noticeable by ebuild maintainers. I might agree to remove colors in other deprecation warnings, but this patch wasn't adding any other deprecation warnings. Not using colors in deprecation warnings doesn't have any benefits and only decreases noticeability of deprecation warnings. -- Arfrever Frehtes Taifersar Arahesis signature.asc Description: This is a digitally signed message part.
Re: [gentoo-dev] Minor changes in python.eclass and distutils.eclass
On Monday, July 05, 2010 16:23:50 Arfrever Frehtes Taifersar Arahesis wrote: 2010-07-05 21:18:57 Mark Loeser napisał(a): Arfrever Frehtes Taifersar Arahesis said: 2010-07-05 20:00:11 Mark Loeser napisał(a): Everyone else has already made valid points. I'm just picking this one to reply to now. Please remove the colors you have added. If you need a new function, say eqawarn, we should have that added in the next EAPI with a description of when and where to use it. In case of the colored message added in this patch, if einfo/elog/ewarn/eqawarn/eerror was used, then its output wouldn't be logged by Portage. I don't understand what you are trying to say. Portage doesn't log output of einfo/elog/ewarn/eqawarn/eerror called in global scope. dont call it in global scope then. the more parsing you do in global scope the worse you make performance for the tree. these things get executed during dependency generation which means it gets spammed even when not emerging. use pkg_setup or something similar like everyone else. -mike signature.asc Description: This is a digitally signed message part.
Re: [gentoo-dev] Minor changes in python.eclass and distutils.eclass
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 05-07-2010 15:23, Arfrever Frehtes Taifersar Arahesis wrote: These minor changes in python.eclass and distutils.eclass have been already reviewed on alias of Gentoo Python Project. It's recommended to be familiar with internals of current code before trying to understand these minor changes. Suggestions about indentation and quoting will be rejected. Arfrever, I'm not going to delve into the details that have been addressed all other this thread. Instead I'll just address one small issue. The use of *minor* in the title of this thread and the sheer size of the patch attached are not compatible. Please don't label changes such as these as *minor* in future instances. - -- Regards, Jorge Vicetto (jmbsvicetto) - jmbsvicetto at gentoo dot org Gentoo- forums / Userrel / Devrel / KDE / Elections -BEGIN PGP SIGNATURE- Version: GnuPG v2.0.15 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ iQIcBAEBAgAGBQJMMmHQAAoJEC8ZTXQF1qEPzG4P/RMaVOxp/cCTq5Y5RZQq0IIE t6nh8AJGl0nJyjDi4LtpI2A1BpC2m0EcekDmYLkQouo9+IRwIrbawI4N8MQFAiZC Nj/yideygJMFYE7o6RiIxZ3HtCxhoV0uY48c0azxehwVBykrLKx2GNLLvmveqGcj 85saWl0awP9/KktYYUt+EHTBsTmJCmQX9493pxWUrMczFrz+eyDMjgDc9La77SuU u3JEFLLmvHtLgY8rAoSm3zDulxsc+fmdWTsS1/36Ko+StAQLMDz1OGUEsokQPSMB IeouOKxHGIlefaAYhrqtIni+M9hkxFRLS46zmcrD9J36SYcPrD0XagWtEdb31+Ar PPssQJnUz4FaVXd4rLVhBdOc2FpZR1VqaMK2wVhYKm1aq+We68LGxEd5JU0PCPJT Y9Do3YfolXyfsF/zDkmuq0v2613i7APqVrmfK3kMHn0u8Ytma1JdLD/KzF/AOwTM WCoH5Q5os59oN0gpbYUx1EHzIq0qfY6IslZLu/cedN6hM9QK+TOWzks2bqa888iI J6h9kd20FcfddAbbBIfhe0jSk1kRxVYV7RLfvsBZ2FB7PeIKqli+wDBUOj36FmU3 i+W5v5Uu134Dpdn+pevtFeDZsu4bmXrkasg8HvlH7J2oquCu2Egb88ST0dQEB08Z lNYugouUn/oPIrgk76t5 =qBaG -END PGP SIGNATURE-
Re: [gentoo-dev] Minor changes in python.eclass and distutils.eclass
On Mon, 05 Jul 2010 22:50:56 + Jorge Manuel B. S. Vicetto jmbsvice...@gentoo.org wrote: I'm not going to delve into the details that have been addressed all other this thread. Instead I'll just address one small issue. The use of *minor* in the title of this thread and the sheer size of the patch attached are not compatible. Please don't label changes such as these as *minor* in future instances. Another major flaw on its own is the sheer size of the patch (it hadn't been explicitly mentioned to my knowledge). Aren't we supposed to commit early and often in open source projects instead of dumping all your changes on reviewers at once (first fork and later merge)? And if you're still going to do it, a the huge patch should only be necessary to establish a single sweeping change, not many small changes. jer
Re: [gentoo-dev] Minor changes in python.eclass and distutils.eclass
2010-07-06 00:29:09 Mike Frysinger napisał(a): On Monday, July 05, 2010 16:23:50 Arfrever Frehtes Taifersar Arahesis wrote: 2010-07-05 21:18:57 Mark Loeser napisał(a): Arfrever Frehtes Taifersar Arahesis said: 2010-07-05 20:00:11 Mark Loeser napisał(a): Everyone else has already made valid points. I'm just picking this one to reply to now. Please remove the colors you have added. If you need a new function, say eqawarn, we should have that added in the next EAPI with a description of when and where to use it. In case of the colored message added in this patch, if einfo/elog/ewarn/eqawarn/eerror was used, then its output wouldn't be logged by Portage. I don't understand what you are trying to say. Portage doesn't log output of einfo/elog/ewarn/eqawarn/eerror called in global scope. dont call it in global scope then. the more parsing you do in global scope the worse you make performance for the tree. these things get executed during dependency generation which means it gets spammed even when not emerging. use pkg_setup or something similar like everyone else. Only 84 ebuilds in the tree set NEED_PYTHON variable. NEED_PYTHON can be used only in EAPI =2. python_pkg_setup() is optional in EAPI =3. Exactly 0 ebuilds, which use NEED_PYTHON, also call python_pkg_setup(). EXPORT_FUNCTIONS wasn't ommitted during this calculation. There are over 27000 ebuilds in the tree. Performance penalty from =84 ebuilds shouldn't be noticeable during `emerge --regen`. -- Arfrever Frehtes Taifersar Arahesis signature.asc Description: This is a digitally signed message part.
Re: [gentoo-dev] Minor changes in python.eclass and distutils.eclass
On Monday, July 05, 2010 19:31:59 Arfrever Frehtes Taifersar Arahesis wrote: Only 84 ebuilds in the tree set NEED_PYTHON variable. NEED_PYTHON can be used only in EAPI =2. python_pkg_setup() is optional in EAPI =3. Exactly 0 ebuilds, which use NEED_PYTHON, also call python_pkg_setup(). EXPORT_FUNCTIONS wasn't ommitted during this calculation. well then, it shouldnt be hard for you to fix that small number and drop this crap from the tree. your global scope echo still doesnt require color support. everyone is telling you for obvious reasons to stop doing this. so do it already. There are over 27000 ebuilds in the tree. Performance penalty from =84 ebuilds shouldn't be noticeable during `emerge --regen`. your few ebuilds and my few ebuilds and that guys few ebuilds and ... see how it all adds up ? -mike signature.asc Description: This is a digitally signed message part.
