[gentoo-dev] Re: [PATCH 09/15] perl-module.eclass: Use DIST_ prefix instead of MODULE_ prefix in EAPI=6 for control variables.

[Duncan]

Andreas K. Huettel posted on Sat, 12 Dec 2015 21:29:30 +0100 as excerpted:

> Am Samstag, 12. Dezember 2015, 10:15:05 schrieb Duncan:
>> 
>> What I would have expected to see here (and in other docs-related
>> patches) would be cleanly separate EAPI-5 vs. EAPI-6 descriptions, on
>> separate lines, so when EAPI-5 support cleanup time comes, it's easy
>> to simply delete EAPI-5 lines.  Something like:
>> 
> Well... In a way that certainly makes sense. However, ...
> 
> Once this is out I would personally prefer all bumps to be also EAPI
> bumps, and all new perl-module.eclass ebuilds to use EAPI=6. Adding
> extensive documentation about EAPI=5 seems to be counterproductive
> there.

What I was (perhaps too) gently hinting was that docs aren't just for 
devs.  Users use them too, when debugging existing ebuild failures, etc.  
Presumably existing EAPI-5 perl-module.eclass consuming ebuilds will be 
around, at least in stable, for a few months, correct?

But upon closer look I see only functions were previously documented (as 
of my tree from 10 days ago anyway), so you're adding var documentation 
that wasn't there at all earlier and it's not a labeling regression after 
all.  Given that any documentation is better than none...

I understand doing it the way you did it, now. =:^)

Tho I'd still probably put that last EAPI-5 sentence on its own line, 
again, making EAPI-5 cleanup when the time comes that much easier. =:^)

-- 
Duncan - List replies preferred.   No HTML msgs.
"Every nonfree program has a lord, a master --
and if you use the program, he is your master."  Richard Stallman

[gentoo-portage-dev] [PATCH] _dep_check_composite_db._visible: verify that highest_visible matches (bug 567686)

[Zac Medico]

If the highest visible match for a package slot does not match the
required atom, then do not mask other packages in the same slot.
Bug 567686 was triggered when the highest visible match for the
package slot did not match the subslot specified by the required atom.

X-Gentoo-Bug: 567686
X-Gentoo-Bug-url: https://bugs.gentoo.org/show_bug.cgi?id=567686
---
 pym/_emerge/depgraph.py | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/pym/_emerge/depgraph.py b/pym/_emerge/depgraph.py
index 2169b00..fd2c771 100644
--- a/pym/_emerge/depgraph.py
+++ b/pym/_emerge/depgraph.py
@@ -9064,7 +9064,9 @@ class _dep_check_composite_db(dbapi):
# Note: highest_visible is not necessarily the real 
highest
# visible, especially when --update is not enabled, so 
use
# < operator instead of !=.
-   if highest_visible is not None and pkg < 
highest_visible:
+   if (highest_visible is not None and pkg < 
highest_visible
+   and atom_set.findAtomForPackage(highest_visible,
+   
modified_use=self._depgraph._pkg_use_enabled(highest_visible))):
return False
elif in_graph != pkg:
# Mask choices for packages that would trigger a slot
-- 
2.4.10

Re: [gentoo-dev] News for Gentoo CI and pull request CI

[Pacho Ramos]

El vie, 11-12-2015 a las 23:13 +0100, Michał Górny escribió:
> Hi, everyone.
> 
> I'd like to make a short announcement that in the last few days I've
> been using my little free time to hack on my CI scripts, and they had
> a few major improvements. Most notably:
> 
> 1. pkgcheck now checks for outdated Manifests. In other words,
> gentoo-ci will most likely complain about missing or extraneous
> Manifest entries before Infra does, and pull request checks should
> complain loudly about them before the PRs are merged.
> 
> 2. The scripts use XML output of pkgcheck and group issues by
> packages,
> rather than grepping text output for failure indications
> and considering successive lines a single failure. Which means better
> output, better permalinks, better navigation and a completely new
> ugly
> UI.
> 
> 3. Most of the persisting issues (like deprecations) are skipped now,
> and all the output fits on one page. And that page is usually easier
> on
> the browser than each of the splits were.
> 
> 4. gentoo-ci compares previous check results correctly, and splits
> the output into three groups: new issues, previous issues and fixes
> issues.
> 
> 5. gentoo-ci bisects on new issues, pinpoints the exact commit
> introducing the failure and CCs both the author and the committer.
> In other words, does the hard work for me and does it more
> efficiently.
> It also has nice bisect caching support which makes bisecting a
> number
> of issues caused by the same commit very fast.
> 
> 6. pull-request checks now distinguish between issues introduced by
> PR
> and carried over from ::gentoo. Not that you should ignore the latter
> entirely since they may hide additional issues from PR.
> 
> Most of this fun stuff came from a simple obvious thing that only
> recently occurred to me: I don't have to run a full pkgcheck for
> bisecting! I just need to check the packages that are failing
> in the most recent commit!
> 
> Of course, the whole thing is open source:
> 
> 1. https://bitbucket.org/mgorny/repo-mirror-ci
> 
> 2. https://bitbucket.org/mgorny/pkgcheck-result-parser
> 
> Enjoy!
> 

Thanks a lot!

Re: [gentoo-dev] [PATCH 09/15] perl-module.eclass: Use DIST_ prefix instead of MODULE_ prefix in EAPI=6 for control variables.

[Michał Górny]

Dnia 11 grudnia 2015 22:03:05 CET, dilfri...@gentoo.org napisał(a):
>From: "Andreas K. Huettel (dilfridge)" 
>
>This is more consistent with upstream terminology and avoids clashes in
>case of kernel modules.
>---
>eclass/perl-module.eclass | 52
>+++
> 1 file changed, 30 insertions(+), 22 deletions(-)
>
>diff --git a/eclass/perl-module.eclass b/eclass/perl-module.eclass
>index 82707fe..efcc47c 100644
>--- a/eclass/perl-module.eclass
>+++ b/eclass/perl-module.eclass
>@@ -114,30 +114,31 @@ esac
> 
> LICENSE="${LICENSE:-|| ( Artistic GPL-1+ )}"
> 
>-# @ECLASS-VARIABLE: MODULE_NAME
>+# @ECLASS-VARIABLE: DIST_NAME
> # @DESCRIPTION:
># (EAPI=6) This variable provides a way to override PN for the
>calculation of S,
> # SRC_URI, and HOMEPAGE. Defaults to PN.
> 
>-# @ECLASS-VARIABLE: MODULE_VERSION
>+# @ECLASS-VARIABLE: DIST_VERSION
> # @DESCRIPTION:
>-# This variable provides a way to override PV for the calculation of S
>and SRC_URI.
>+# (EAPI=6) This variable provides a way to override PV for the
>calculation of S and SRC_URI.
># Use it to provide the non-normalized, upstream version number.
>Defaults to PV.
>+# Named MODULE_VERSION in EAPI=5.
> 
>-# @ECLASS-VARIABLE: MODULE_A_EXT
>+# @ECLASS-VARIABLE: DIST_A_EXT
> # @DESCRIPTION:
>-# This variable provides a way to override the distfile extension for
>the calculation of
>-# SRC_URI. Defaults to tar.gz
>+# (EAPI=6) This variable provides a way to override the distfile
>extension for the calculation of
>+# SRC_URI. Defaults to tar.gz. Named MODULE_A_EXT in EAPI=5.
> 
>-# @ECLASS-VARIABLE: MODULE_A
>+# @ECLASS-VARIABLE: DIST_A
> # @DESCRIPTION:
>-# This variable provides a way to override the distfile name for the
>calculation of
>-# SRC_URI. Defaults to
>${MODULE_NAME}-${MODULE_VERSION}.${MODULE_A_EXT}
>+# (EAPI=6) This variable provides a way to override the distfile name
>for the calculation of
>+# SRC_URI. Defaults to ${DIST_NAME}-${DIST_VERSION}.${DIST_A_EXT}
>Named MODULE_A in EAPI=5.
> 
>-# @ECLASS-VARIABLE: MODULE_AUTHOR
>+# @ECLASS-VARIABLE: DIST_AUTHOR
> # @DESCRIPTION:
>-# This variable sets the module author name for the calculation of
>-# SRC_URI.
>+# (EAPI=6) This variable sets the module author name for the
>calculation of
>+# SRC_URI. Named MODULE_AUTHOR in EAPI=5.
> 
> if [[ ${EAPI:-0} = 5 ]] ; then
>   if [[ -n ${MY_PN} || -n ${MY_PV} || -n ${MODULE_VERSION} ]] ; then
>@@ -146,19 +147,26 @@ if [[ ${EAPI:-0} = 5 ]] ; then
>   fi
>   MODULE_NAME=${MY_PN:-${PN}}
>   MODULE_P=${MY_P:-${P}}
>+
>+  [[ -z "${SRC_URI}" && -z "${MODULE_A}" ]] && \
>+  MODULE_A="${MODULE_P}.${MODULE_A_EXT:-tar.gz}"
>+  [[ -z "${SRC_URI}" && -n "${MODULE_AUTHOR}" ]] && \
>+  
>SRC_URI="mirror://cpan/authors/id/${MODULE_AUTHOR:0:1}/${MODULE_AUTHOR:0:2}/${MODULE_AUTHOR}/${MODULE_SECTION:+${MODULE_SECTION}/}${MODULE_A}"
>+  [[ -z "${HOMEPAGE}" ]] && \
>+  HOMEPAGE="http://search.cpan.org/dist/${MODULE_NAME}/;
> else
>-  MODULE_NAME=${MODULE_NAME:-${PN}}
>-  MODULE_P=${MODULE_NAME}-${MODULE_VERSION:-${PV}}
>-  S=${WORKDIR}/${MODULE_P}
>+  DIST_NAME=${DIST_NAME:-${PN}}
>+  DIST_P=${DIST_NAME}-${DIST_VERSION:-${PV}}
>+  S=${WORKDIR}/${DIST_P}
>+
>+  [[ -z "${SRC_URI}" && -z "${DIST_A}" ]] && \
>+  DIST_A="${DIST_P}.${DIST_A_EXT:-tar.gz}"
>+  [[ -z "${SRC_URI}" && -n "${DIST_AUTHOR}" ]] && \
>+  
>SRC_URI="mirror://cpan/authors/id/${DIST_AUTHOR:0:1}/${DIST_AUTHOR:0:2}/${DIST_AUTHOR}/${DIST_SECTION:+${DIST_SECTION}/}${DIST_A}"
>+  [[ -z "${HOMEPAGE}" ]] && \
>+  HOMEPAGE="http://search.cpan.org/dist/${DIST_NAME}/;
> fi
> 
>-[[ -z "${SRC_URI}" && -z "${MODULE_A}" ]] && \
>-  MODULE_A="${MODULE_P}.${MODULE_A_EXT:-tar.gz}"
>-[[ -z "${SRC_URI}" && -n "${MODULE_AUTHOR}" ]] && \
>-  
>SRC_URI="mirror://cpan/authors/id/${MODULE_AUTHOR:0:1}/${MODULE_AUTHOR:0:2}/${MODULE_AUTHOR}/${MODULE_SECTION:+${MODULE_SECTION}/}${MODULE_A}"
>-[[ -z "${HOMEPAGE}" ]] && \
>-  HOMEPAGE="http://search.cpan.org/dist/${MODULE_NAME}/;
>-
> SRC_PREP="no"
> SRC_TEST="skip"
> PREFER_BUILDPL="yes"

Now that's confusing history. You introduce new variables, document them all, 
then suddenly rename then. Wouldn't it be better to rename old ones first and 
introduce new ones directly with correct names?
-- 
Sent from my Android device with K-9 Mail. Please excuse my brevity.

Re: [gentoo-dev] [PATCH 10/15] perl-module.eclass: Rename SRC_TEST to DIST_TEST in EAPI=6 and default to "do parallel"

[Michał Górny]

Dnia 11 grudnia 2015 22:03:06 CET, dilfri...@gentoo.org napisał(a):
>From: "Andreas K. Huettel (dilfridge)" 
>
>---
>eclass/perl-module.eclass | 40 +++-
> 1 file changed, 23 insertions(+), 17 deletions(-)
>
>diff --git a/eclass/perl-module.eclass b/eclass/perl-module.eclass
>index efcc47c..0d428d2 100644
>--- a/eclass/perl-module.eclass
>+++ b/eclass/perl-module.eclass
>@@ -154,6 +154,8 @@ if [[ ${EAPI:-0} = 5 ]] ; then
>   
> SRC_URI="mirror://cpan/authors/id/${MODULE_AUTHOR:0:1}/${MODULE_AUTHOR:0:2}/${MODULE_AUTHOR}/${MODULE_SECTION:+${MODULE_SECTION}/}${MODULE_A}"
>   [[ -z "${HOMEPAGE}" ]] && \
>   HOMEPAGE="http://search.cpan.org/dist/${MODULE_NAME}/;
>+
>+  SRC_TEST="skip"
> else
>   DIST_NAME=${DIST_NAME:-${PN}}
>   DIST_P=${DIST_NAME}-${DIST_VERSION:-${PV}}
>@@ -168,7 +170,6 @@ else
> fi
> 
> SRC_PREP="no"
>-SRC_TEST="skip"
> PREFER_BUILDPL="yes"
> 
> pm_echovar=""
>@@ -315,6 +316,18 @@ perl-module_src_compile() {
>   fi
> }
> 
>+# @ECLASS-VARIABLE: DIST_TEST
>+# @DESCRIPTION:
>+# (EAPI=6) Variable that controls if tests are run in the test phase
>+# at all, and if yes under which conditions. Defaults to "do parallel"
>+# In EAPI=5 the variable is called SRC_TEST and defaults to "skip".
>+# All of the following have been tested to work:
>+#  DIST_TEST="do parallel"
>+#  DIST_TEST="parallel"
>+#  DIST_TEST="parallel do"
>+#  DIST_TEST=parallel
>+#  DIST_TEST=skip

At this point, I'm starting to wonder why you aren't using PERL_ prefix...

>+
> # @FUNCTION: perl-module_src-test
> # @USAGE: perl-module_src_test()
> # @DESCRIPTION:
>@@ -323,24 +336,17 @@ perl-module_src_compile() {
> #
> # If you want more verbose testing, set TEST_VERBOSE=1
> # in your bashrc | /etc/portage/make.conf | ENV
>-#
>-# or ebuild writers:
>-# If you wish to enable default tests w/ 'make test' ,
>-#
>-#  SRC_TEST="do"
>-#
>-# If you wish to have threads run in parallel ( using the users
>makeopts )
>-# all of the following have been tested to work.
>-#
>-#  SRC_TEST="do parallel"
>-#  SRC_TEST="parallel"
>-#  SRC_TEST="parallel do"
>-#  SRC_TEST=parallel
>-#
> perl-module_src_test() {
>   debug-print-function $FUNCNAME "$@"
>-  if has 'do' ${SRC_TEST} || has 'parallel' ${SRC_TEST} ; then
>-  if has "${TEST_VERBOSE:-0}" 0 && has 'parallel' ${SRC_TEST} ; 
>then
>+  local my_test_control
>+  if [[ ${EAPI:-0} = 5 ]] ; then
>+  my_test_control=${SRC_TEST}
>+  else
>+  my_test_control=${DIST_TEST:-do parallel}
>+  fi
>+
>+  if has 'do' ${my_test_control} || has 'parallel' ${my_test_control} ;
>then
>+  if has "${TEST_VERBOSE:-0}" 0 && has 'parallel' 
>${my_test_control} ;
>then
>   export HARNESS_OPTIONS=j$(makeopts_jobs)
>   einfo "Test::Harness Jobs=$(makeopts_jobs)"
>   fi

-- 
Sent from my Android device with K-9 Mail. Please excuse my brevity.

Re: [gentoo-dev] [PATCH 01/15] perl-module.eclass: Allow EAPI=6, ban PERL_EXPORT_PHASE_FUNCTIONS there, simplify perl dependency

[Michał Górny]

Dnia 11 grudnia 2015 22:02:57 CET, dilfri...@gentoo.org napisał(a):
>From: "Andreas K. Huettel (dilfridge)" 
>
>---
>eclass/perl-module.eclass | 54
>++-
> 1 file changed, 44 insertions(+), 10 deletions(-)
>
>diff --git a/eclass/perl-module.eclass b/eclass/perl-module.eclass
>index 5a476d2..82c72fc 100644
>--- a/eclass/perl-module.eclass
>+++ b/eclass/perl-module.eclass
>@@ -37,21 +37,55 @@ case "${EAPI:-0}" in
>   RDEPEND="${DEPEND}"
>   ;;
>   esac
>+
>+  case "${PERL_EXPORT_PHASE_FUNCTIONS:-yes}" in
>+  yes)
>+  EXPORT_FUNCTIONS ${PERL_EXPF}
>+  ;;
>+  no)
>+  debug-print "PERL_EXPORT_PHASE_FUNCTIONS=no"
>+  ;;
>+  *)
>+  die 
>"PERL_EXPORT_PHASE_FUNCTIONS=${PERL_EXPORT_PHASE_FUNCTIONS} is
>not supported by perl-module.eclass"
>+  ;;
>+  esac

You could split this move, you know.

>   ;;
>-  *)
>-  die "EAPI=${EAPI} is not supported by perl-module.eclass"
>-  ;;
>-esac
>+  6)
>+  [[ ${CATEGORY} == "perl-core" ]] && \
>+  PERL_EXPF+=" pkg_postinst pkg_postrm"
>+
>+  case "${GENTOO_DEPEND_ON_PERL:-yes}" in
>+  yes)
>+  case "${GENTOO_DEPEND_ON_PERL_SUBSLOT:-yes}" in
>+  yes)
>+  DEPEND="dev-lang/perl:="
>+  ;;
>+  *)
>+  DEPEND="dev-lang/perl"
>+  ;;
>+  esac
>+  RDEPEND="${DEPEND}"
>+  ;;
>+  esac

Why don't you pack this stuff into a single trinary variable?

>+
>+  case "${PERL_EXPORT_PHASE_FUNCTIONS:-undefined}" in

The ':' means it's not going to complain about PERL_EXPORT_PHASE_FUNCTIONS=.

>+  yes)
>+  die "PERL_EXPORT_PHASE_FUNCTIONS is banned in 
>EAPI=6. It defaults
>to yes anyway."
>+  ;;
>+  no)
>+  die "PERL_EXPORT_PHASE_FUNCTIONS is banned in 
>EAPI=6. Inherit
>perl-functions.eclass instead."
>+  ;;
>+  undefined)
>+  ;;
>+  *)
>+  die "PERL_EXPORT_PHASE_FUNCTIONS is banned in 
>EAPI=6."
>+  ;;
>+  esac
> 
>-case "${PERL_EXPORT_PHASE_FUNCTIONS:-yes}" in
>-  yes)
>   EXPORT_FUNCTIONS ${PERL_EXPF}
>   ;;
>-  no)
>-  debug-print "PERL_EXPORT_PHASE_FUNCTIONS=no"
>-  ;;
>   *)
>-  die "PERL_EXPORT_PHASE_FUNCTIONS=${PERL_EXPORT_PHASE_FUNCTIONS} 
>is
>not supported by perl-module.eclass"
>+  die "EAPI=${EAPI} is not supported by perl-module.eclass"
>   ;;
> esac
> 

-- 
Sent from my Android device with K-9 Mail. Please excuse my brevity.

[gentoo-portage-dev] How do we want to deal with Arfrever's bunch of patches?

[Alexander Berntsen]

-BEGIN PGP SIGNED MESSAGE-
Hash: SHA512

Resending this from an address that will actually reach the list...

Friends,

As I'm sure you've noticed, Arfrever pushed a bunch of unreviewed
stuff. Michał reverted some of them as they were breaking changes in
addition to being unreviewed (oh, and of course they don't follow the
commit message format -- why would they). There's a bunch left.

Do we want to cherry-pick Zac's patches (i.e. revert Arfrever's
patches), and have him put them up for review as usual? It's a bit of
a mess because of Zac's patches. If we go down this route it would
likely be easiest for Zac to revert them, since he'll be the more
confident regarding what to keep in the ensuing merge conflicts.

Arfrever asked that someone review the patches and keep them if they
are OK, and revert only the ones that are problematic. I think this
would be OK for now, but obviously not something we want to encourage
or allow in the future.

I will not sanction a public history rewrite, but in general I'm open
to other suggestions.
- -- 
Alexander
alexan...@plaimi.net
https://secure.plaimi.net/~alexander
- -- 
Alexander
berna...@gentoo.org
https://secure.plaimi.net/~alexander
-BEGIN PGP SIGNATURE-
Version: GnuPG v2
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=M0Pq
-END PGP SIGNATURE-

Re: [gentoo-dev] [PATCH 05/15] perl-module.eclass: Use eapply from EAPI=6 on and always call eapply_user

[Michał Górny]

Dnia 11 grudnia 2015 22:03:01 CET, dilfri...@gentoo.org napisał(a):
>From: "Andreas K. Huettel (dilfridge)" 
>
>---
> eclass/perl-module.eclass | 14 +++---
> 1 file changed, 11 insertions(+), 3 deletions(-)
>
>diff --git a/eclass/perl-module.eclass b/eclass/perl-module.eclass
>index 8caa981..9d87b17 100644
>--- a/eclass/perl-module.eclass
>+++ b/eclass/perl-module.eclass
>@@ -138,9 +138,17 @@ perl-module_src_unpack() {
> # This function is to be called during the ebuild src_prepare() phase.
> perl-module_src_prepare() {
>   debug-print-function $FUNCNAME "$@"
>-  [[ ${PATCHES[@]} ]] && epatch "${PATCHES[@]}"
>-  debug-print "$FUNCNAME: applying user patches"
>-  epatch_user
>+
>+  if [[ ${EAPI:-0} = 5 ]] ; then
>+  [[ ${PATCHES[@]} ]] && epatch "${PATCHES[@]}"
>+  debug-print "$FUNCNAME: applying user patches"
>+  epatch_user
>+  else
>+  [[ ${PATCHES[@]} ]] && eapply "${PATCHES[@]}"
>+  debug-print "$FUNCNAME: applying user patches"
>+  eapply_user

Why not just 'default'?

>+  fi
>+
>   if [[ ${PERL_RM_FILES[@]} ]]; then
>   debug-print "$FUNCNAME: stripping unneeded files"
>   perl_rm_files "${PERL_RM_FILES[@]}"

-- 
Sent from my Android device with K-9 Mail. Please excuse my brevity.

Re: [gentoo-dev] News for Gentoo CI and pull request CI

[Patrice Clement]

Friday 11 Dec 2015 23:13:33, Michał Górny wrote :
> Hi, everyone.
> 
> I'd like to make a short announcement that in the last few days I've
> been using my little free time to hack on my CI scripts, and they had
> a few major improvements. Most notably:
> 
> 1. pkgcheck now checks for outdated Manifests. In other words,
> gentoo-ci will most likely complain about missing or extraneous
> Manifest entries before Infra does, and pull request checks should
> complain loudly about them before the PRs are merged.
> 
> 2. The scripts use XML output of pkgcheck and group issues by packages,
> rather than grepping text output for failure indications
> and considering successive lines a single failure. Which means better
> output, better permalinks, better navigation and a completely new ugly
> UI.
> 
> 3. Most of the persisting issues (like deprecations) are skipped now,
> and all the output fits on one page. And that page is usually easier on
> the browser than each of the splits were.
> 
> 4. gentoo-ci compares previous check results correctly, and splits
> the output into three groups: new issues, previous issues and fixes
> issues.
> 
> 5. gentoo-ci bisects on new issues, pinpoints the exact commit
> introducing the failure and CCs both the author and the committer.
> In other words, does the hard work for me and does it more efficiently.
> It also has nice bisect caching support which makes bisecting a number
> of issues caused by the same commit very fast.
> 
> 6. pull-request checks now distinguish between issues introduced by PR
> and carried over from ::gentoo. Not that you should ignore the latter
> entirely since they may hide additional issues from PR.
> 
> Most of this fun stuff came from a simple obvious thing that only
> recently occurred to me: I don't have to run a full pkgcheck for
> bisecting! I just need to check the packages that are failing
> in the most recent commit!
> 
> Of course, the whole thing is open source:
> 
> 1. https://bitbucket.org/mgorny/repo-mirror-ci
> 
> 2. https://bitbucket.org/mgorny/pkgcheck-result-parser
> 
> Enjoy!
> 
> -- 
> Best regards,
> Michał Górny
> 


Awesome Michal, thanks a lot!

-- 
Patrice Clement
Gentoo Linux developer
http://www.gentoo.org


signature.asc
Description: PGP signature

Re: [gentoo-dev] [PATCH 03/15] perl-module.eclass: Just go ahead when calling configure twice in EAPI=6

[Michał Górny]

Dnia 11 grudnia 2015 22:02:59 CET, dilfri...@gentoo.org napisał(a):
>From: "Andreas K. Huettel (dilfridge)" 
>
>---
> eclass/perl-module.eclass | 4 +++-
> 1 file changed, 3 insertions(+), 1 deletion(-)
>
>diff --git a/eclass/perl-module.eclass b/eclass/perl-module.eclass
>index 78fa01d..734de11 100644
>--- a/eclass/perl-module.eclass
>+++ b/eclass/perl-module.eclass
>@@ -146,7 +146,9 @@ perl-module_src_prepare() {
> perl-module_src_configure() {
>   debug-print-function $FUNCNAME "$@"
> 
>-  [[ ${SRC_PREP} = yes ]] && return 0
>+  if [[ ${EAPI:-0} = 5 ]]; then
>+  [[ ${SRC_PREP} = yes ]] && return 0
>+  fi

Why not combine this into single [[ ]]?

>   SRC_PREP="yes"
> 
>   perl_check_env

-- 
Sent from my Android device with K-9 Mail. Please excuse my brevity.

Re: [gentoo-dev] [PATCH 07/15] perl-module.eclass: Do not do any magic with MY_... variables in EAPI=6 anymore

[Michał Górny]

Dnia 11 grudnia 2015 22:03:03 CET, dilfri...@gentoo.org napisał(a):
>From: Andreas K. Hüttel 
>
>---
> eclass/perl-module.eclass | 16 +++-
> 1 file changed, 11 insertions(+), 5 deletions(-)
>
>diff --git a/eclass/perl-module.eclass b/eclass/perl-module.eclass
>index 6a5f2ff..7434bcf 100644
>--- a/eclass/perl-module.eclass
>+++ b/eclass/perl-module.eclass
>@@ -102,12 +102,18 @@ esac
> 
> LICENSE="${LICENSE:-|| ( Artistic GPL-1+ )}"
> 
>-if [[ -n ${MY_PN} || -n ${MY_PV} || -n ${MODULE_VERSION} ]] ; then
>-  : ${MY_P:=${MY_PN:-${PN}}-${MY_PV:-${MODULE_VERSION:-${PV
>-  S=${MY_S:-${WORKDIR}/${MY_P}}
>+if [[ ${EAPI:-0} = 5 ]] ; then
>+  if [[ -n ${MY_PN} || -n ${MY_PV} || -n ${MODULE_VERSION} ]] ; then
>+  : ${MY_P:=${MY_PN:-${PN}}-${MY_PV:-${MODULE_VERSION:-${PV
>+  S=${MY_S:-${WORKDIR}/${MY_P}}
>+  fi
>+  MODULE_NAME=${MY_PN:-${PN}}
>+  MODULE_P=${MY_P:-${P}}
>+else
>+  MODULE_NAME=${MODULE_NAME:-${PN}}
>+  MODULE_P=${MODULE_NAME}-${MODULE_VERSION:-${PV}}
>+  S=${WORKDIR}/${MODULE_P}
> fi
>-MODULE_NAME=${MY_PN:-${PN}}
>-MODULE_P=${MY_P:-${P}}
> 
> [[ -z "${SRC_URI}" && -z "${MODULE_A}" ]] && \
>   MODULE_A="${MODULE_P}.${MODULE_A_EXT:-tar.gz}"

How likely is it that one would use MY_* without MODULE_*? Maybe it would be 
good to warn about this?
-- 
Sent from my Android device with K-9 Mail. Please excuse my brevity.

Re: [gentoo-dev] [PATCH 06/15] perl-module.eclass: Introduce MODULE_NAME and MODULE_P while still keeping existing weirdness compatible

[Michał Górny]

Dnia 11 grudnia 2015 22:03:02 CET, dilfri...@gentoo.org napisał(a):
>From: Andreas K. Hüttel 
>
>---
> eclass/perl-module.eclass | 6 --
> 1 file changed, 4 insertions(+), 2 deletions(-)
>
>diff --git a/eclass/perl-module.eclass b/eclass/perl-module.eclass
>index 9d87b17..6a5f2ff 100644
>--- a/eclass/perl-module.eclass
>+++ b/eclass/perl-module.eclass
>@@ -106,13 +106,15 @@ if [[ -n ${MY_PN} || -n ${MY_PV} || -n
>${MODULE_VERSION} ]] ; then
>   : ${MY_P:=${MY_PN:-${PN}}-${MY_PV:-${MODULE_VERSION:-${PV
>   S=${MY_S:-${WORKDIR}/${MY_P}}
> fi
>+MODULE_NAME=${MY_PN:-${PN}}
>+MODULE_P=${MY_P:-${P}}
> 
> [[ -z "${SRC_URI}" && -z "${MODULE_A}" ]] && \
>-  MODULE_A="${MY_P:-${P}}.${MODULE_A_EXT:-tar.gz}"
>+  MODULE_A="${MODULE_P}.${MODULE_A_EXT:-tar.gz}"
> [[ -z "${SRC_URI}" && -n "${MODULE_AUTHOR}" ]] && \
>   
> SRC_URI="mirror://cpan/authors/id/${MODULE_AUTHOR:0:1}/${MODULE_AUTHOR:0:2}/${MODULE_AUTHOR}/${MODULE_SECTION:+${MODULE_SECTION}/}${MODULE_A}"
> [[ -z "${HOMEPAGE}" ]] && \
>-  HOMEPAGE="http://search.cpan.org/dist/${MY_PN:-${PN}}/;
>+  HOMEPAGE="http://search.cpan.org/dist/${MODULE_NAME}/;
> 
> SRC_PREP="no"
> SRC_TEST="skip"

Here, the commit message summary line is way too long.


-- 
Sent from my Android device with K-9 Mail. Please excuse my brevity.

Re: [gentoo-dev] News item: Python ABIFLAGS rebuild needed

[Ulrich Mueller]

> On Fri, 11 Dec 2015, Mike Gilbert wrote:

> It is worth noting that GLEP 42 makes no mention of EAPI, and
> current versions of portage do support slots in dependency atoms
> used in news items.

> I guess it's a matter of documenting this somewhere. Where would
> that be?

We should update GLEP 42 and introduce a News-Item-Format 1.1 that
supports EAPI 5 for dependency syntax.

(This has come up several times in the past already, but for some
reason nobody is willing to take the initiative for updating the GLEP.
Everyone seems to be happy with a workaround for their own news item.)

Ulrich


pgpNJKZ8QzIGx.pgp
Description: PGP signature

[gentoo-dev] Re: [PATCH 09/15] perl-module.eclass: Use DIST_ prefix instead of MODULE_ prefix in EAPI=6 for control variables.

[Duncan]

dilfridge posted on Fri, 11 Dec 2015 22:03:05 +0100 as excerpted:

> -# @ECLASS-VARIABLE: MODULE_VERSION
> +# @ECLASS-VARIABLE: DIST_VERSION
>  # @DESCRIPTION:
> -# This variable provides a way to override PV for the calculation of S and 
> SRC_URI.
> +# (EAPI=6) This variable provides a way to override PV for the calculation 
> of S and SRC_URI.
>  # Use it to provide the non-normalized, upstream version number. Defaults to 
> PV.
> +# Named MODULE_VERSION in EAPI=5.

Dev-author's discretion, here and below, as it's purely style, but...

What I would have expected to see here (and in other docs-related patches) 
would be cleanly
separate EAPI-5 vs. EAPI-6 descriptions, on separate lines, so when EAPI-5 
support cleanup
time comes, it's easy to simply delete EAPI-5 lines.  Something like:

+# ECLASS-VARIABLE: DIST_VERSION
+# (EAPI=6)
+# This variable provides a way to override PV for the calculation of S and 
SRC_URI.
+
 # @ECLASS-VARIABLE: MODULE_VERSION
 # @DESCRIPTION:
-# This variable provides a way to override PV for the calculation of S and 
SRC_URI.
+# (EAPI-5 version of DIST_VERSION)

With this, EAPI-5 cleanup should be purely lines deleted (including the EAPI-6 
on its own line).
Additionally, people (like me) trying to debug failing existing EAPI-5 ebuilds 
won't be having
to look for EAPI-5 var-names at unexpected locations in the documentation. =:^)

Of course for the other vars that change names but not functionality between 
the two
EAPIs, both in this patch and in others, as well.

(Don't miss the further comment below.)

> @@ -146,19 +147,26 @@ if [[ ${EAPI:-0} = 5 ]] ; then
>   fi
>   MODULE_NAME=${MY_PN:-${PN}}
>   MODULE_P=${MY_P:-${P}}
> +
> + [[ -z "${SRC_URI}" && -z "${MODULE_A}" ]] && \
> + MODULE_A="${MODULE_P}.${MODULE_A_EXT:-tar.gz}"
> + [[ -z "${SRC_URI}" && -n "${MODULE_AUTHOR}" ]] && \
> + 
> SRC_URI="mirror://cpan/authors/id/${MODULE_AUTHOR:0:1}/${MODULE_AUTHOR:0:2}/${MODULE_AUTHOR}/${MODULE_SECTION:+${MODULE_SECTION}/}${MODULE_A}"
> + [[ -z "${HOMEPAGE}" ]] && \
> + HOMEPAGE="http://search.cpan.org/dist/${MODULE_NAME}/;
>  else
> - MODULE_NAME=${MODULE_NAME:-${PN}}
> - MODULE_P=${MODULE_NAME}-${MODULE_VERSION:-${PV}}
> - S=${WORKDIR}/${MODULE_P}
> + DIST_NAME=${DIST_NAME:-${PN}}
> + DIST_P=${DIST_NAME}-${DIST_VERSION:-${PV}}
> + S=${WORKDIR}/${DIST_P}
> +
> + [[ -z "${SRC_URI}" && -z "${DIST_A}" ]] && \
> + DIST_A="${DIST_P}.${DIST_A_EXT:-tar.gz}"
> + [[ -z "${SRC_URI}" && -n "${DIST_AUTHOR}" ]] && \
> + 
> SRC_URI="mirror://cpan/authors/id/${DIST_AUTHOR:0:1}/${DIST_AUTHOR:0:2}/${DIST_AUTHOR}/${DIST_SECTION:+${DIST_SECTION}/}${DIST_A}"
> + [[ -z "${HOMEPAGE}" ]] && \
> + HOMEPAGE="http://search.cpan.org/dist/${DIST_NAME}/;
>  fi
>  
> -[[ -z "${SRC_URI}" && -z "${MODULE_A}" ]] && \
> - MODULE_A="${MODULE_P}.${MODULE_A_EXT:-tar.gz}"
> -[[ -z "${SRC_URI}" && -n "${MODULE_AUTHOR}" ]] && \
> - 
> SRC_URI="mirror://cpan/authors/id/${MODULE_AUTHOR:0:1}/${MODULE_AUTHOR:0:2}/${MODULE_AUTHOR}/${MODULE_SECTION:+${MODULE_SECTION}/}${MODULE_A}"
> -[[ -z "${HOMEPAGE}" ]] && \
> - HOMEPAGE="http://search.cpan.org/dist/${MODULE_NAME}/;
> -

Particularly for 100% straight var-renames with no change in functionality,
I'd expect something more like (pseudocode):

if $EAPI = 5 then
# eapi-6+ only, clear them
unset newvarname newvarname2

# assign eapi-6 vars from eapi-5 vars so we
# can reference only the new eapi-6 ones below
newvarname=oldvarname
newvarname2=oldvarname2
...
fi

# make sure we don't use old eapi-5 varnames
unset oldvarname1 oldvarname2

# the rest of the patch simply changes old varnames to new ones,
# thus avoiding duplicate code


That makes it very clear in both the patch and the post-patch code
what's going on, while preventing all three of:
1) use of the old varnames in EAPI-6 (due to the unset)
2) use of the new varnames in EAPI-5 (due to the unset and reassignment)
3) uncaught and unchanged reference to the old vars (due to the unset)

Again, when eapi-5 code cleanup time comes, it's just deletion of the
eapi-5 lines.  No else clause to unwind.  And in the mean time, the code
isn't duplicated to eapi-5 and eapi-6 locations with only differing vars.
=:^)

Seems cleaner to me, but as I said above, purely a matter of author style,
so if you don't like it, feel free to ignore.

-- 
Duncan - List replies preferred.   No HTML msgs.
"Every nonfree program has a lord, a master --
and if you use the program, he is your master."  Richard Stallman

Re: [gentoo-dev] [PATCH 11/15] perl-module.eclass: Introduce DIST_TEST_OVERRIDE in EAPI=6

[Michał Górny]

Dnia 11 grudnia 2015 22:03:07 CET, dilfri...@gentoo.org napisał(a):
>From: "Andreas K. Huettel (dilfridge)" 
>
>---
> eclass/perl-module.eclass | 11 ++-
> 1 file changed, 10 insertions(+), 1 deletion(-)
>
>diff --git a/eclass/perl-module.eclass b/eclass/perl-module.eclass
>index 0d428d2..4dc39eb 100644
>--- a/eclass/perl-module.eclass
>+++ b/eclass/perl-module.eclass
>@@ -328,6 +328,14 @@ perl-module_src_compile() {
> #  DIST_TEST=parallel
> #  DIST_TEST=skip
> 
>+# @ECLASS-VARIABLE: DIST_TEST_OVERRIDE
>+# @DESCRIPTION:
>+# (EAPI=6) Variable that controls if tests are run in the test phase
>+# at all, and if yes under which conditions. It is intended for use in
>+# make.conf or the environment by ebuild authors during testing, and
>+# accepts the same values as DIST_TEST. If set, it overrides DIST_TEST
>+# completely. DO NOT USE THIS IN EBUILDS!
>+
> # @FUNCTION: perl-module_src-test
> # @USAGE: perl-module_src_test()
> # @DESCRIPTION:
>@@ -342,7 +350,8 @@ perl-module_src_test() {
>   if [[ ${EAPI:-0} = 5 ]] ; then
>   my_test_control=${SRC_TEST}
>   else
>-  my_test_control=${DIST_TEST:-do parallel}
>+  [[ -n "${DIST_TEST_OVERRIDE}" ]] && ewarn DIST_TEST_OVERRIDE is 
>set
>to ${DIST_TEST_OVERRIDE}
>+  my_test_control=${DIST_TEST_OVERRIDE:-${DIST_TEST:-do parallel}}
>   fi
> 
>   if has 'do' ${my_test_control} || has 'parallel' ${my_test_control} ;
>then

Why that? This seems really redundant to RESTRICT=test.

-- 
Sent from my Android device with K-9 Mail. Please excuse my brevity.

Re: [gentoo-dev] [PATCH 11/15] perl-module.eclass: Introduce DIST_TEST_OVERRIDE in EAPI=6

[Michał Górny]

Dnia 12 grudnia 2015 13:05:12 CET, "Michał Górny"  
napisał(a):
>
>
>Dnia 11 grudnia 2015 22:03:07 CET, dilfri...@gentoo.org napisał(a):
>>From: "Andreas K. Huettel (dilfridge)" 
>>
>>---
>> eclass/perl-module.eclass | 11 ++-
>> 1 file changed, 10 insertions(+), 1 deletion(-)
>>
>>diff --git a/eclass/perl-module.eclass b/eclass/perl-module.eclass
>>index 0d428d2..4dc39eb 100644
>>--- a/eclass/perl-module.eclass
>>+++ b/eclass/perl-module.eclass
>>@@ -328,6 +328,14 @@ perl-module_src_compile() {
>> #  DIST_TEST=parallel
>> #  DIST_TEST=skip
>> 
>>+# @ECLASS-VARIABLE: DIST_TEST_OVERRIDE
>>+# @DESCRIPTION:
>>+# (EAPI=6) Variable that controls if tests are run in the test phase
>>+# at all, and if yes under which conditions. It is intended for use
>in
>>+# make.conf or the environment by ebuild authors during testing, and
>>+# accepts the same values as DIST_TEST. If set, it overrides
>DIST_TEST
>>+# completely. DO NOT USE THIS IN EBUILDS!
>>+
>> # @FUNCTION: perl-module_src-test
>> # @USAGE: perl-module_src_test()
>> # @DESCRIPTION:
>>@@ -342,7 +350,8 @@ perl-module_src_test() {
>>  if [[ ${EAPI:-0} = 5 ]] ; then
>>  my_test_control=${SRC_TEST}
>>  else
>>- my_test_control=${DIST_TEST:-do parallel}
>>+ [[ -n "${DIST_TEST_OVERRIDE}" ]] && ewarn DIST_TEST_OVERRIDE is 
>>set
>>to ${DIST_TEST_OVERRIDE}
>>+ my_test_control=${DIST_TEST_OVERRIDE:-${DIST_TEST:-do parallel}}
>>  fi
>> 
>>  if has 'do' ${my_test_control} || has 'parallel' ${my_test_control} ;
>>then
>
>Why that? This seems really redundant to RESTRICT=test.

I meant FEATURES=-test. Sorry for the confusion.

-- 
Sent from my Android device with K-9 Mail. Please excuse my brevity.

Re: [gentoo-portage-dev] How do we want to deal with Arfrever's bunch of patches?

[Michał Górny]

On Sat, 12 Dec 2015 13:01:04 +0100
Alexander Berntsen  wrote:

> Friends,
> 
> As I'm sure you've noticed, Arfrever pushed a bunch of unreviewed
> stuff. Michał reverted some of them as they were breaking changes in
> addition to being unreviewed (oh, and of course they don't follow the
> commit message format -- why would they). There's a bunch left.
> 
> Do we want to cherry-pick Zac's patches (i.e. revert Arfrever's
> patches), and have him put them up for review as usual? It's a bit of
> a mess because of Zac's patches. If we go down this route it would
> likely be easiest for Zac to revert them, since he'll be the more
> confident regarding what to keep in the ensuing merge conflicts.
> 
> Arfrever asked that someone review the patches and keep them if they
> are OK, and revert only the ones that are problematic. I think this
> would be OK for now, but obviously not something we want to encourage
> or allow in the future.

Here's my quick review of the remaining commits (newest first):


31923f4 portage.package.ebuild.config.config.__init__(): Skip some warnings for 
Portage Python scripts called in ebuild environment.
7853950 Delete support for PORTDIR and PORTDIR_OVERLAY from make.conf and 
environment.
e7d95cb portage.repository.config.RepoConfig: Support location with trailing 
whitespace by using quoting.

Those three already reverted.


8036223 _emerge.depgraph.depgraph._select_files(): Update message.

Can probably be left, it's updating messages for repos.conf which is still 
valid.


9716f8a emirrordist: Delete support for deprecated --portdir and 
--portdir-overlay options.
2d40eb4 egencache: Delete support for deprecated --portdir and 
--portdir-overlay options.

Not sure. Maybe something relies on this.


fb4d1f4 ebuild: Rename some variables.

I'd keep it, doesn't hurt anything and the new names are more readable than 
'myrepo' ;-).


9e104c4 ebuild: Set PORTAGE_REPOSITORIES instead of deprecated PORTDIR_OVERLAY.

Kill it as relying on weirdo PORTAGE_REPOSITORIES invention.


3917544 bin/socks5-server.py: Fix DeprecationWarning with Python >=3.4.4.

This one looks good, and even the commit message is sane.


2cde1f6 portage.repository.config: Clean reading of repository names and drop 
support for repositories with missing or invalid names.

I'd revert it, it's a breaking change.


1064765 portage.repository.config.RepoConfig: Delete user_location attribute 
and consistently use location attribute everywhere.

I think we can keep this if it doesn't break anything. I didn't hear
about this attribute before, so I suggest Zac reviews this.


48c2af4 Respect PYTHONDONTWRITEBYTECODE in test suite.

I don't mind either way. But can't we make this simpler?


528dc7c portage.package.ebuild.fetch.fetch(): Clean setting of variables.

Purely stylistic change.


10cccf7 Support environmental variables overriding parts of configuration of 
repositories.

Kill this ugly thing with a lot of fire. This is so wrong you can't get
much worse.


39d81c5 portage.package.ebuild.config.config.__getitem__(): Partially drop 
backward compatibility for nonexistent keys.

Maybe keep it but needs someone smarter than me to review.


4f25fa6 ebuild: Move imports to the top.
5ea789a 
portage.package.ebuild._config.LicenseManager.LicenseManager._getPkgAcceptLicense():
 Fix typo in call to hasattr().

Makes sense.


3ffdbbe ebuild: Do not catch unexpected KeyErrors from aux_get().

Maybe. Need others to confirm.


b8da04b portageq envvar: Return 1 for any nonexistent variable.

Makes sense.


53a0b63 _emerge.actions.getgccversion(): Work with chost=None and with no 
explicit chost.

Probably makes sense, needs extra review.


628aa10 portage.sync.modules.cvs.CheckCVSConfig.check_cvs_repo(): Fix 
"KeyError: 'sync-cvs-repo'".

Makes sense.


f05056c portage.repository.config.RepoConfigLoader._parse(): Delete unused 
portdir parametre.
041fc70 portage.repository.config.RepoConfigLoader._add_repositories(): Delete 
unused ignored_location_map parametre.
c1a2714 portage.repository.config.RepoConfigLoader._parse(): Delete unused 
ignored_map and ignored_location_map parametres.

Those all make sense.

-- 
Best regards,
Michał Górny



pgpvMFEuGVeiX.pgp
Description: OpenPGP digital signature

Re: [gentoo-dev] [PATCH 11/15] perl-module.eclass: Introduce DIST_TEST_OVERRIDE in EAPI=6

[Andreas K. Huettel]

-BEGIN PGP SIGNED MESSAGE-
Hash: SHA512

Am Samstag, 12. Dezember 2015, 13:06:59 schrieb Michał Górny:
> Dnia 12 grudnia 2015 13:05:12 CET, "Michał Górny"  
napisał(a):
> >Dnia 11 grudnia 2015 22:03:07 CET, dilfri...@gentoo.org napisał(a):
> >>From: "Andreas K. Huettel (dilfridge)" 
> >
> >Why that? This seems really redundant to RESTRICT=test.
> 
> I meant FEATURES=-test. Sorry for the confusion.

Well, it's less redundant than it is now. :)

Right now SRC_TEST can be used for two things: 
1) disabling tests (indeed redundantly redundant)
2) and running tests non-parallel or parallelized

The new version, DIST_TEST and DIST_TEST_OVERRIDE, allows a bit more 
finegrained control. 
- - verbose or not
- - enable network tests or not
- - parallelized or not

As Perl team member I'd likely set something like
DIST_TEST_OVERRIDE="do parallel" 
in my make.conf... which means that even ebuilds where the parallel testing 
caused problems in the past and has been disabled in the ebuild try to do 
that...

- -- 

Andreas K. Huettel
Gentoo Linux developer 
dilfri...@gentoo.org
http://www.akhuettel.de/

-BEGIN PGP SIGNATURE-
Version: GnuPG v2.1
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=7aVq
-END PGP SIGNATURE-

Re: [gentoo-dev] [PATCH 03/15] perl-module.eclass: Just go ahead when calling configure twice in EAPI=6

[Andreas K. Huettel]

-BEGIN PGP SIGNED MESSAGE-
Hash: SHA512

Am Samstag, 12. Dezember 2015, 12:50:29 schrieb Michał Górny:
> >+if [[ ${EAPI:-0} = 5 ]]; then
> >+[[ ${SRC_PREP} = yes ]] && return 0
> >+fi
> 
> Why not combine this into single [[ ]]?
> 

yep, better!


- -- 

Andreas K. Huettel
Gentoo Linux developer 
dilfri...@gentoo.org
http://www.akhuettel.de/

-BEGIN PGP SIGNATURE-
Version: GnuPG v2.1
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=F5DS
-END PGP SIGNATURE-

Re: [gentoo-dev] News item: Python ABIFLAGS rebuild needed

[Mike Gilbert]

On Sat, Dec 12, 2015 at 3:54 AM, Ulrich Mueller  wrote:
>> On Fri, 11 Dec 2015, Mike Gilbert wrote:
>
>> It is worth noting that GLEP 42 makes no mention of EAPI, and
>> current versions of portage do support slots in dependency atoms
>> used in news items.
>
>> I guess it's a matter of documenting this somewhere. Where would
>> that be?
>
> We should update GLEP 42 and introduce a News-Item-Format 1.1 that
> supports EAPI 5 for dependency syntax.
>
> (This has come up several times in the past already, but for some
> reason nobody is willing to take the initiative for updating the GLEP.
> Everyone seems to be happy with a workaround for their own news item.)

Thank you. After skimming thought GLEP 1 to refresh my memory, I have
filed bug 568068 to propose the change.

https://bugs.gentoo.org/show_bug.cgi?id=568068

Re: [gentoo-dev] Re: converting copyright/license information in OpenRC

[William Hubbs]

On Fri, Dec 11, 2015 at 02:16:30PM -0800, Patrick McLean wrote:
> On Fri, 11 Dec 2015 15:37:48 -0600
> William Hubbs  wrote:
> 
> > On Fri, Dec 11, 2015 at 09:04:47PM +0100, Ulrich Mueller wrote:
> > > > On Fri, 11 Dec 2015, William Hubbs wrote:  
> > >   
> > Well, the OpenRC project is currently inconsistent about this, so the
> > intention is to make it consistent.
> > 
> >  The .c/.h files have file-scope licenses, but that isn't true for
> >  everything in the project.
> > 
> > I am willing to make the effort to do this, I was just wondering if
> > there are any legal pitfalls I need to worry about.
> > 
> > My theory is I can probably use git to find out who all of the authors
> > are, and generate an Authors list from that information and from
> > looking at copyright notices.
> > 
> 
> One concern about this is the possibility of copied code. If OpenRC
> ever copied code from other BSD licensed projects, then dropping the
> notice from the top of the file would be a violation of the upstream
> license.

The only thing I know about immediately that we have copied from an
upstream is queue.h, and sure, I wasn't planning on touching that file.
The same applies to any other files copied from an upstream that I find,
or any functions that I see that are obviously copied. I'm thinking I
will be able to detect this by extra copyright notices that are from
sources not listed in the log.

William



signature.asc
Description: Digital signature

Re: [gentoo-dev] [PATCH 09/15] perl-module.eclass: Use DIST_ prefix instead of MODULE_ prefix in EAPI=6 for control variables.

[Andreas K. Huettel]

-BEGIN PGP SIGNED MESSAGE-
Hash: SHA512

Am Samstag, 12. Dezember 2015, 13:01:58 schrieb Michał Górny:
> Dnia 11 grudnia 2015 22:03:05 CET, dilfri...@gentoo.org napisał(a):
> 
> Now that's confusing history. You introduce new variables, document them
> all, then suddenly rename then. Wouldn't it be better to rename old ones
> first and introduce new ones directly with correct names?

Yes. This is what I meant with squashing commits. 

Basically for now I just kept things along my line of thought, but some of 
these single patches can be consolidated once the finetuning is done.

- -- 

Andreas K. Huettel
Gentoo Linux developer 
dilfri...@gentoo.org
http://www.akhuettel.de/

-BEGIN PGP SIGNATURE-
Version: GnuPG v2.1

iQJ8BAEBCgBmBQJWbIQNXxSAAC4AKGlzc3Vlci1mcHJAbm90YXRpb25zLm9w
ZW5wZ3AuZmlmdGhob3JzZW1hbi5uZXQ0RkJDMzI0NjNBOTIwMDY5MTQ2NkMzNDBF
MTM4NkZEN0VGNEI1Nzc5AAoJEOE4b9fvS1d5kTwP/A2/8Xro7h/NApFUOEx90Uz4
XMaRk0bh9XtzLB03Wait4ozxV06qOuNDB9iC99uZMZB4plqmegybPatotaaRlDhA
1StUJk0qfU8KYR3D2EHTsHZb80R9AYWNDwKSnK3c2VbHV0zPxoB0DSTL1ncCyhho
MGwzHOxVXx34o3fNFTaOtm3N0aIIz5eKmqQ5HQpH+CoZrSiFQqYeK3gOANkQCmuc
4j/FYl7yzdTReqifELzmBc62qUFjZET4QBEhq4tKSHPtOc6uJSXmUbwvRj8t3suG
CGZeWvq8i1CspaQrjIpfhz6MyHjxZ7tVLdC94RkM2OH9brwGO6TUl6lEZKOEDz9J
jNZf30sxnCpQ6ICnM0olCMnPdpOcfSTWhzl7PSJu3hIT5nMAbgh7PHry/Rz0T2nw
eP7izpSK6d1j3djf3Z0W9CQkrNjSohEa4xX6PIj/UFM8xHQ5F0ajihmMGb9gTdFI
Gh/JX0c1Q+vM8+9H3oSRvYNgwH1sOCeCBDYFdFwHxeDGNSWufo4xmhIrqJ1YTY81
n87vGc/QLB1d/f92WzDMDNI/WqN0odd+6rYpYE6SFAERZ+XrQXcN1uukbMzMAazd
qHgeLmMpdWxoe3RQygGdh5/xFFc6PzdrDo5OrpuUtMcE4mKGzrPjwohobDU2QhsW
SYOuaMjXnSCc9+EpO31O
=CLg4
-END PGP SIGNATURE-

Re: [gentoo-dev] [PATCH 07/15] perl-module.eclass: Do not do any magic with MY_... variables in EAPI=6 anymore

[Andreas K. Huettel]

-BEGIN PGP SIGNED MESSAGE-
Hash: SHA512

Am Samstag, 12. Dezember 2015, 12:59:08 schrieb Michał Górny:
> Dnia 11 grudnia 2015 22:03:03 CET, dilfri...@gentoo.org napisał(a):
> >
> >-if [[ -n ${MY_PN} || -n ${MY_PV} || -n ${MODULE_VERSION} ]] ; then
> >-: ${MY_P:=${MY_PN:-${PN}}-${MY_PV:-${MODULE_VERSION:-${PV
> >-S=${MY_S:-${WORKDIR}/${MY_P}}
> >+if [[ ${EAPI:-0} = 5 ]] ; then
> >+if [[ -n ${MY_PN} || -n ${MY_PV} || -n ${MODULE_VERSION} ]] ; then
> >+: ${MY_P:=${MY_PN:-${PN}}-${MY_PV:-${MODULE_VERSION:-${PV
> >+S=${MY_S:-${WORKDIR}/${MY_P}}
> >+fi
> >+MODULE_NAME=${MY_PN:-${PN}}
> >+MODULE_P=${MY_P:-${P}}
> >+else
> >+MODULE_NAME=${MODULE_NAME:-${PN}}
> >+MODULE_P=${MODULE_NAME}-${MODULE_VERSION:-${PV}}
> >+S=${WORKDIR}/${MODULE_P}

> How likely is it that one would use MY_* without MODULE_*? Maybe it would
> be good to warn about this?

Well... the main usage of this input is calculation of SRC_URI and S. 

So, I'm not so worried about it, since failing to update the settings will 
lead to problems already when running ebuild manifest...

- -- 

Andreas K. Huettel
Gentoo Linux developer 
dilfri...@gentoo.org
http://www.akhuettel.de/

-BEGIN PGP SIGNATURE-
Version: GnuPG v2.1
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=7sm6
-END PGP SIGNATURE-

Re: [gentoo-portage-dev] How do we want to deal with Arfrever's bunch of patches?

[Zac Medico]

On 12/12/2015 06:50 AM, Michał Górny wrote:
> On Sat, 12 Dec 2015 13:01:04 +0100
> Alexander Berntsen  wrote:
> 
>> Friends,
>>
>> As I'm sure you've noticed, Arfrever pushed a bunch of unreviewed
>> stuff. Michał reverted some of them as they were breaking changes in
>> addition to being unreviewed (oh, and of course they don't follow the
>> commit message format -- why would they). There's a bunch left.
>>
>> Do we want to cherry-pick Zac's patches (i.e. revert Arfrever's
>> patches), and have him put them up for review as usual? It's a bit of
>> a mess because of Zac's patches. If we go down this route it would
>> likely be easiest for Zac to revert them, since he'll be the more
>> confident regarding what to keep in the ensuing merge conflicts.

Yeah, I'd be happy to do that.

>> Arfrever asked that someone review the patches and keep them if they
>> are OK, and revert only the ones that are problematic. I think this
>> would be OK for now, but obviously not something we want to encourage
>> or allow in the future.
> 
> Here's my quick review of the remaining commits (newest first):

Thanks!

> 31923f4 portage.package.ebuild.config.config.__init__(): Skip some warnings 
> for Portage Python scripts called in ebuild environment.
> 7853950 Delete support for PORTDIR and PORTDIR_OVERLAY from make.conf and 
> environment.

Before dropping PORTDIR and PORTDIR_OVERLAY, we need deprecation
warnings for a couple of years in advance.

> e7d95cb portage.repository.config.RepoConfig: Support location with trailing 
> whitespace by using quoting.

This one might be okay. Did it really break anything?

> Those three already reverted.
> 
> 
> 8036223 _emerge.depgraph.depgraph._select_files(): Update message.
> 
> Can probably be left, it's updating messages for repos.conf which is still 
> valid.

Looks good.

> 9716f8a emirrordist: Delete support for deprecated --portdir and 
> --portdir-overlay options.
> 2d40eb4 egencache: Delete support for deprecated --portdir and 
> --portdir-overlay options.
> 
> Not sure. Maybe something relies on this.

I don't care either way. Note that these options have triggered
deprecation warnings since before portage-2.2.0:

https://gitweb.gentoo.org/proj/portage.git/commit/?id=92fa6a1e1fcc92f2ad8ce8896c4ce3ec39477485
https://gitweb.gentoo.org/proj/portage.git/commit/?id=ab7d606eb6f66004a63c9406cf9cf87b352f6d01

> fb4d1f4 ebuild: Rename some variables.
> 
> I'd keep it, doesn't hurt anything and the new names are more readable than 
> 'myrepo' ;-).

Looks good.

> 9e104c4 ebuild: Set PORTAGE_REPOSITORIES instead of deprecated 
> PORTDIR_OVERLAY.
> 
> Kill it as relying on weirdo PORTAGE_REPOSITORIES invention.

I think this one is mostly okay, since PORTAGE_REPOSITORIES has been
supported for some time, and it would be nice to eliminate internal
usage for PORTDIR_OVERLAY.

However, this patch relies on the _read_repo_name method added in
2cde1f6. When we revert 2cde1f6, we can also change this code to use the
RepoConfig._read_valid_repo_name staticmethod that was removed in 2cde1f6.

> 3917544 bin/socks5-server.py: Fix DeprecationWarning with Python >=3.4.4.
> 
> This one looks good, and even the commit message is sane.

Looks good.

> 2cde1f6 portage.repository.config: Clean reading of repository names and drop 
> support for repositories with missing or invalid names.
> 
> I'd revert it, it's a breaking change.

I think this is too much at once, so we should revert it.

As a first step in the direction that this patch is going, we could
remove the special case for /usr/local/portage from the repo_name_check
functions, so that people will start getting warnings for a missing
repo_name there.

> 1064765 portage.repository.config.RepoConfig: Delete user_location attribute 
> and consistently use location attribute everywhere.
> 
> I think we can keep this if it doesn't break anything. I didn't hear
> about this attribute before, so I suggest Zac reviews this.

repo.user_location is the original location specified in repos.conf, and
repo.location == os.path.realpath(repo.user_location)

So, repo.user_location can differ if the path contains any symlinks.

Is there any value in hiding the symlink indirection in repository paths
shown in output like emerge --info? If not, then the change is fine as
long as there are no external consumers of this API. I think there is
very low probability of there being external consumers.

> 48c2af4 Respect PYTHONDONTWRITEBYTECODE in test suite.
> 
> I don't mind either way. But can't we make this simpler?

Looks good.

We can avoid code duplication if we add an environment setup function
for all the tests to share, but that can come in a later patch.

> 528dc7c portage.package.ebuild.fetch.fetch(): Clean setting of variables.
> 
> Purely stylistic change.

I don't mind either way.

> 10cccf7 Support environmental variables overriding parts of configuration of 
> repositories.
> 
> Kill this ugly thing with a lot of fire. This is so 

Re: [gentoo-dev] [PATCH 04/15] perl-module.eclass: Use default unpacking from EAPI=6 on

[Andreas K. Huettel]

-BEGIN PGP SIGNED MESSAGE-
Hash: SHA512

Am Samstag, 12. Dezember 2015, 00:40:52 schrieb Ulrich Mueller:
> > On Fri, 11 Dec 2015, dilfridge  wrote:
> 
> > +   [[ ${EAPI}==5 ]] || die "perl-module_src_unpack is banned in EAPI=6 or
> > later"
> 
> Without spaces around the operator, this condition will always
> evaluate to true. ;-)

Ok will fix this. Reminds me why I keep getting puzzled by Bash... :D

- -- 

Andreas K. Huettel
Gentoo Linux developer 
dilfri...@gentoo.org
http://www.akhuettel.de/

-BEGIN PGP SIGNATURE-
Version: GnuPG v2.1
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=zczv
-END PGP SIGNATURE-

Re: [gentoo-dev] [PATCH 01/15] perl-module.eclass: Allow EAPI=6, ban PERL_EXPORT_PHASE_FUNCTIONS there, simplify perl dependency

[Andreas K. Huettel]

-BEGIN PGP SIGNED MESSAGE-
Hash: SHA512

Am Samstag, 12. Dezember 2015, 12:49:09 schrieb Michał Górny:
> Dnia 11 grudnia 2015 22:02:57 CET, dilfri...@gentoo.org napisał(a):

> >+case "${PERL_EXPORT_PHASE_FUNCTIONS:-yes}" in
> >+yes)
> >+EXPORT_FUNCTIONS ${PERL_EXPF}
> >+;;
> >+no)
> >+debug-print "PERL_EXPORT_PHASE_FUNCTIONS=no"
> >+;;
> >+*)
> >+die 
"PERL_EXPORT_PHASE_FUNCTIONS=${PERL_EXPORT_PHASE_FUNCTIONS} is
> >not supported by perl-module.eclass"
> >+;;
> >+esac
> 
> You could split this move, you know.

Yeah but it's probably more useful to change behavior now and then keep it 
consistent for whole EAPI=6... 

> >+case "${GENTOO_DEPEND_ON_PERL:-yes}" in
> >+yes)
> >+case "${GENTOO_DEPEND_ON_PERL_SUBSLOT:-yes}" in
> >+yes)
> >+DEPEND="dev-lang/perl:="
> >+;;
> >+*)
> >+DEPEND="dev-lang/perl"
> >+;;
> >+esac
> >+RDEPEND="${DEPEND}"
> >+;;
> >+esac
> 
> Why don't you pack this stuff into a single trinary variable?

Good idea. Let's kill GENTOO_DEPEND_ON_PERL_SUBSLOT (never liked it, too long 
variable name :) and instead provide support for
GENTOO_DEPEND_ON_PERL=noslotop


- -- 

Andreas K. Huettel
Gentoo Linux developer 
dilfri...@gentoo.org
http://www.akhuettel.de/

-BEGIN PGP SIGNATURE-
Version: GnuPG v2.1

iQJ8BAEBCgBmBQJWbIHoXxSAAC4AKGlzc3Vlci1mcHJAbm90YXRpb25zLm9w
ZW5wZ3AuZmlmdGhob3JzZW1hbi5uZXQ0RkJDMzI0NjNBOTIwMDY5MTQ2NkMzNDBF
MTM4NkZEN0VGNEI1Nzc5AAoJEOE4b9fvS1d5/NEP/RshKssht2YpkvxMzhj1wIj2
6ZAIWpq7lUlgaaORy/wVeGu29GVG1g5+Ov2KZdZjUVtVdG5JcRkdxihgwWL+AUT9
VfvEnlikEN+UTjJFE+1YmDEAynPpfAjv31UBLQg1LiKaUCHE8kABUFCJrUfmf6Pl
A0CkiwdGCDjeY3BFdQG4wQm/rbnMSeyMK39qodqNfKk/vwDl67LSHvifBKzuHqcJ
xcgKHVPNgN5T0HR6GLiBRO83PWOyQVYyX0/rMkQqTW4ggKGMHHtj15e+193sAyzX
ke1+yrYQtA1nLa3bwSRz/gWJkxXEPo9ZsRxPYgk1F3wM/l63hGYBQ5dR7oR2hx69
hoAhmwQvv4GpJoDq5Tm8//Ko+joRcFrj1yNTNmQn/KSWlX7+PE8uMrh2IoVznXb+
7BhWpeIx5/6txwlfGZzvdkH0KdK+Ahv0+QsKF0ZHm3S8T2fWsFLYfYR2cuC1UQYz
1W5vWaCUyt8nBZU7VR/+MWzyffZlwGkPmckbAxDX5CSn5EJTVs+y0PyhduYKd6RN
9i+8bmNHxPmehpjrs2En2iOhHiwO6q30wUTh0zY6CmNWmSsywUPlF8HSYwwScF0y
9C326aVegA29465gNdUw8njknwkksyUzFRBviGaB/ZzhjuA7AIvvdq/lpa3Eakhl
DXfw03GJDBK9cJogc2x+
=Zio8
-END PGP SIGNATURE-

Re: [gentoo-dev] [PATCH 05/15] perl-module.eclass: Use eapply from EAPI=6 on and always call eapply_user

[Andreas K. Huettel]

-BEGIN PGP SIGNED MESSAGE-
Hash: SHA512

Am Samstag, 12. Dezember 2015, 12:54:35 schrieb Michał Górny:
> Dnia 11 grudnia 2015 22:03:01 CET, dilfri...@gentoo.org napisał(a):
> >diff --git a/eclass/perl-module.eclass b/eclass/perl-module.eclass
> >index 8caa981..9d87b17 100644
> >--- a/eclass/perl-module.eclass
> >+++ b/eclass/perl-module.eclass
> >@@ -138,9 +138,17 @@ perl-module_src_unpack() {
> >
> > # This function is to be called during the ebuild src_prepare() phase.
> > perl-module_src_prepare() {
> > 
> > debug-print-function $FUNCNAME "$@"
> >
> >-[[ ${PATCHES[@]} ]] && epatch "${PATCHES[@]}"
> >-debug-print "$FUNCNAME: applying user patches"
> >-epatch_user
> >+
> >+if [[ ${EAPI:-0} = 5 ]] ; then
> >+[[ ${PATCHES[@]} ]] && epatch "${PATCHES[@]}"
> >+debug-print "$FUNCNAME: applying user patches"
> >+epatch_user
> >+else
> >+[[ ${PATCHES[@]} ]] && eapply "${PATCHES[@]}"
> >+debug-print "$FUNCNAME: applying user patches"
> >+eapply_user
> 
> Why not just 'default'?
> 

Much better, yes.

- -- 

Andreas K. Huettel
Gentoo Linux developer 
dilfri...@gentoo.org
http://www.akhuettel.de/

-BEGIN PGP SIGNATURE-
Version: GnuPG v2.1
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=E1if
-END PGP SIGNATURE-

Re: [gentoo-dev] Re: [PATCH 09/15] perl-module.eclass: Use DIST_ prefix instead of MODULE_ prefix in EAPI=6 for control variables.

[Andreas K. Huettel]

-BEGIN PGP SIGNED MESSAGE-
Hash: SHA512

Am Samstag, 12. Dezember 2015, 10:15:05 schrieb Duncan:
> 
> What I would have expected to see here (and in other docs-related patches)
> would be cleanly separate EAPI-5 vs. EAPI-6 descriptions, on separate
> lines, so when EAPI-5 support cleanup time comes, it's easy to simply
> delete EAPI-5 lines.  Something like:
> 

Well... In a way that certainly makes sense. However, ...

Once this is out I would personally prefer all bumps to be also EAPI bumps, 
and all new perl-module.eclass ebuilds to use EAPI=6. Adding extensive 
documentation about EAPI=5 seems to be counterproductive there.

- -- 

Andreas K. Huettel
Gentoo Linux developer 
dilfri...@gentoo.org
http://www.akhuettel.de/

-BEGIN PGP SIGNATURE-
Version: GnuPG v2.1
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=jmuY
-END PGP SIGNATURE-

Re: [gentoo-dev] [PATCH 10/15] perl-module.eclass: Rename SRC_TEST to DIST_TEST in EAPI=6 and default to "do parallel"

[Kent Fredric]

On 13 December 2015 at 09:41, Andreas K. Huettel  wrote:
> Well. We had that idea already, and at some point something barfed, I dont
> remember the details anymore. Maybe someone else from the team knows.
>
> The problem is introducing too many variables that look like something
> *upstream* might use...
>
> (We in addition already had creative variable clashes with the kernel module
> build system in the past.)
>
> "Gentoo Perl" would lead to GPL_... as variable prefix, not really suitable.
> Kent suggested GPLX_... which I find horrible. GPERL_? GP_? ...


Yeah. There's a bit of a potential issue with using the PERL_ prefix
in that that prefix is heavily overloaded /in perl/ things, and `perl
-V` subsequently reports all ENV vars with a PERL prefix ( underscore
or not )

PERL_MOM=Yours perl -V  | grep MOM
PERL_MOM="Yours"

So using the PERL_ prefix increases our chances we'll pick something
that will collide, or will mislead somebody into thinking that we're
using an ENV var that Perl reacts magically to, where the ENV var is
purely Gentoo and Purely part of our private API ( At least, its
supposed to be private from the context of a perl interpreter )

If there was a way to mark those vars so they didn't get exported to
perl that would help, but it wouldn't reduce the "What does this do"
impression-confusion.

-- 
Kent

KENTNL - https://metacpan.org/author/KENTNL

[gentoo-portage-dev] [PATCH] SyncManager.sync: always return 4-tuple (bug 567932)

[Zac Medico]

Since commit 496ff326dc18890889d1ea5d2aec590394635960, invalid repos.conf
settings or failure of the pre_sync method caused SyncManager.sync to
return an incorrect number of values.

Fixes: 496ff326dc18 ("sync repositories in parallel (bug 557426)")
X-Gentoo-Bug: 567932
X-Gentoo-Bug-url: https://bugs.gentoo.org/show_bug.cgi?id=567932
---
 pym/portage/sync/controller.py | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/pym/portage/sync/controller.py b/pym/portage/sync/controller.py
index e71ba67..92be3cb 100644
--- a/pym/portage/sync/controller.py
+++ b/pym/portage/sync/controller.py
@@ -129,16 +129,17 @@ class SyncManager(object):
self.repo = repo
self.exitcode = 1
self.updatecache_flg = False
+   hooks_enabled = master_hooks or not 
repo.sync_hooks_only_on_change
if repo.sync_type in self.module_names:
tasks = 
[self.module_controller.get_class(repo.sync_type)]
else:
msg = "\n%s: Sync module '%s' is not an installed/known 
type'\n" \
% (bad("ERROR"), repo.sync_type)
-   return self.exitcode, msg
+   return self.exitcode, msg, self.updatecache_flg, 
hooks_enabled
 
rval = self.pre_sync(repo)
if rval != os.EX_OK:
-   return rval, None
+   return rval, None, self.updatecache_flg, hooks_enabled
 
# need to pass the kwargs dict to the modules
# so they are available if needed.
@@ -157,7 +158,6 @@ class SyncManager(object):
taskmaster = TaskHandler(callback=self.do_callback)
taskmaster.run_tasks(tasks, func, status, options=task_opts)
 
-   hooks_enabled = False
if (master_hooks or self.updatecache_flg or
not repo.sync_hooks_only_on_change):
hooks_enabled = True
-- 
2.4.10

Re: [gentoo-portage-dev] [PATCH] SyncManager.sync: always return 4-tuple (bug 567932)

[Brian Dolbec]

On Sat, 12 Dec 2015 15:09:43 -0800
Zac Medico  wrote:

> Since commit 496ff326dc18890889d1ea5d2aec590394635960, invalid
> repos.conf settings or failure of the pre_sync method caused
> SyncManager.sync to return an incorrect number of values.
> 
> Fixes: 496ff326dc18 ("sync repositories in parallel (bug 557426)")
> X-Gentoo-Bug: 567932
> X-Gentoo-Bug-url: https://bugs.gentoo.org/show_bug.cgi?id=567932
> ---
>  pym/portage/sync/controller.py | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/pym/portage/sync/controller.py
> b/pym/portage/sync/controller.py index e71ba67..92be3cb 100644
> --- a/pym/portage/sync/controller.py
> +++ b/pym/portage/sync/controller.py
> @@ -129,16 +129,17 @@ class SyncManager(object):
>   self.repo = repo
>   self.exitcode = 1
>   self.updatecache_flg = False
> + hooks_enabled = master_hooks or not
> repo.sync_hooks_only_on_change if repo.sync_type in self.module_names:
>   tasks =
> [self.module_controller.get_class(repo.sync_type)] else:
>   msg = "\n%s: Sync module '%s' is not an
> installed/known type'\n" \ % (bad("ERROR"), repo.sync_type)
> - return self.exitcode, msg
> + return self.exitcode, msg,
> self.updatecache_flg, hooks_enabled 
>   rval = self.pre_sync(repo)
>   if rval != os.EX_OK:
> - return rval, None
> + return rval, None, self.updatecache_flg,
> hooks_enabled 
>   # need to pass the kwargs dict to the modules
>   # so they are available if needed.
> @@ -157,7 +158,6 @@ class SyncManager(object):
>   taskmaster = TaskHandler(callback=self.do_callback)
>   taskmaster.run_tasks(tasks, func, status,
> options=task_opts) 
> - hooks_enabled = False
>   if (master_hooks or self.updatecache_flg or
>   not repo.sync_hooks_only_on_change):
>   hooks_enabled = True

yeah, looks good

-- 
Brian Dolbec