[PATCH 1/4] pip_install_wheel: don't lazy assign PIPINSTALLARGS

[PATCH 1/4] pip_install_wheel: don't lazy assign PIPINSTALLARGS

[Ross Burton]

If we expect users to extend this we should use =, as otherwise a recipe
that does += will replace the default value.

Signed-off-by: Ross Burton <ross.burton@arm.com>
---
 meta/classes/pip_install_wheel.bbclass | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/meta/classes/pip_install_wheel.bbclass b/meta/classes/pip_install_wheel.bbclass
index c1680a24ed..febcc8e445 100644
--- a/meta/classes/pip_install_wheel.bbclass
+++ b/meta/classes/pip_install_wheel.bbclass
@@ -10,7 +10,7 @@ PIP_INSTALL_PACKAGE ?= "${@guess_pip_install_package_name(d)}"
 PIP_INSTALL_DIST_PATH ?= "${@d.getVar('SETUPTOOLS_SETUP_PATH') or d.getVar('B')}/dist"
 PYPA_WHEEL ??= "${PIP_INSTALL_DIST_PATH}/${PIP_INSTALL_PACKAGE}-*-*.whl"
 
-PIP_INSTALL_ARGS ?= "\
+PIP_INSTALL_ARGS = "\
     -vvvv \
     --ignore-installed \
     --no-cache \
-- 
2.25.1

[PATCH 2/4] python3-pip: remove obsolete and confusing comment

[Ross Burton]

This comment is very obsolete.

Signed-off-by: Ross Burton <ross.burton@arm.com>
---
 meta/recipes-devtools/python/python3-pip_22.0.3.bb | 1 -
 1 file changed, 1 deletion(-)

diff --git a/meta/recipes-devtools/python/python3-pip_22.0.3.bb b/meta/recipes-devtools/python/python3-pip_22.0.3.bb
index 7eb9fb69ba..8d6dfcede7 100644
--- a/meta/recipes-devtools/python/python3-pip_22.0.3.bb
+++ b/meta/recipes-devtools/python/python3-pip_22.0.3.bb
@@ -45,7 +45,6 @@ EOF
 }
 
 do_install:append() {
-    # Install as pip3 and leave pip2 as default
     if [ -e ${D}/${bindir}/pip ]; then
         rm ${D}/${bindir}/pip
     fi
-- 
2.25.1

[PATCH 3/4] flit_core: clean up configure/compile

[Ross Burton]

Make do_configure do nothing by default as the base configure isn't
useful, and invoke the build API directly instead of writing a script.

In the future, this can simply call python3-build.

Signed-off-by: Ross Burton <ross.burton@arm.com>
---
 meta/classes/flit_core.bbclass | 10 ++++------
 1 file changed, 4 insertions(+), 6 deletions(-)

diff --git a/meta/classes/flit_core.bbclass b/meta/classes/flit_core.bbclass
index 819d12457c..a7667795fe 100644
--- a/meta/classes/flit_core.bbclass
+++ b/meta/classes/flit_core.bbclass
@@ -3,15 +3,13 @@ inherit pip_install_wheel python3native python3-dir setuptools3-base
 DEPENDS += "python3 python3-flit-core-native python3-pip-native"
 
 flit_core_do_configure () {
-    mkdir -p ${S}/dist
-    cat > ${S}/build-it.py << EOF
-from flit_core import buildapi
-buildapi.build_wheel('./dist')
-EOF
+    :
 }
 
+# TODO: ideally this uses pypa/build
 flit_core_do_compile () {
-    ${STAGING_BINDIR_NATIVE}/${PYTHON_PN}-native/${PYTHON_PN} ${S}/build-it.py
+    mkdir -p ${S}/dist
+    nativepython3 -c "from flit_core import buildapi; buildapi.build_wheel('./dist')"
 }
 
 EXPORT_FUNCTIONS do_configure do_compile
-- 
2.25.1

[PATCH 4/4] setuptools_build_meta: clean up configure/compile

[Ross Burton]

Make do_configure do nothing by default as the base configure isn't
useful, and invoke the build API directly instead of writing a script.

In the future, this can simply call python3-build.

Signed-off-by: Ross Burton <ross.burton@arm.com>
---
 meta/classes/setuptools_build_meta.bbclass | 11 ++++-------
 1 file changed, 4 insertions(+), 7 deletions(-)

diff --git a/meta/classes/setuptools_build_meta.bbclass b/meta/classes/setuptools_build_meta.bbclass
index 8f5479e922..0651cca0a0 100644
--- a/meta/classes/setuptools_build_meta.bbclass
+++ b/meta/classes/setuptools_build_meta.bbclass
@@ -3,16 +3,13 @@ inherit pip_install_wheel setuptools3-base
 DEPENDS += "python3-setuptools-native python3-wheel-native"
 
 setuptools_build_meta_do_configure () {
-    mkdir -p ${S}/dist
-    cat > ${S}/build-it.py << EOF
-from setuptools import build_meta
-wheel = build_meta.build_wheel('./dist')
-print(wheel)
-EOF
+    :
 }
 
+# TODO: ideally this uses pypa/build
 setuptools_build_meta_do_compile () {
-    nativepython3 ${S}/build-it.py
+    mkdir -p ${S}/dist
+    nativepython3 -c "from setuptools import build_meta; build_meta.build_wheel('./dist')"
 }
 
 EXPORT_FUNCTIONS do_configure do_compile
-- 
2.25.1

Re: [OE-core] [PATCH 1/4] pip_install_wheel: don't lazy assign PIPINSTALLARGS

[Konrad Weihmann]

On 07.03.22 20:42, Ross Burton wrote:
> If we expect users to extend this we should use =, as otherwise a recipe
> that does += will replace the default value.
> 
> Signed-off-by: Ross Burton <ross.burton@arm.com>
> ---
>   meta/classes/pip_install_wheel.bbclass | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/meta/classes/pip_install_wheel.bbclass b/meta/classes/pip_install_wheel.bbclass
> index c1680a24ed..febcc8e445 100644
> --- a/meta/classes/pip_install_wheel.bbclass
> +++ b/meta/classes/pip_install_wheel.bbclass
> @@ -10,7 +10,7 @@ PIP_INSTALL_PACKAGE ?= "${@guess_pip_install_package_name(d)}"
>   PIP_INSTALL_DIST_PATH ?= "${@d.getVar('SETUPTOOLS_SETUP_PATH') or d.getVar('B')}/dist"
>   PYPA_WHEEL ??= "${PIP_INSTALL_DIST_PATH}/${PIP_INSTALL_PACKAGE}-*-*.whl"
>   
> -PIP_INSTALL_ARGS ?= "\
> +PIP_INSTALL_ARGS = "\
>       -vvvv \
>       --ignore-installed \
>       --no-cache \
> 
> 

Hmm all of the classes use ?= and I think for a reason - this isn't the 
fault of the bbclass but of the recipe in the end, as += is known to 
cause this depending on where the additions have been added (before the 
inherit or after it) - I think the better way would be to use 
PIP_INSTALL_ARGS:append in the actual recipe - guess the fallout would 
be way less than the one I expect from this change.

The big question is if that should changeable by the user or not - I 
personally think it should be

> 
> -=-=-=-=-=-=-=-=-=-=-=-
> Links: You receive all messages sent to this group.
> View/Reply Online (#162860): https://lists.openembedded.org/g/openembedded-core/message/162860
> Mute This Topic: https://lists.openembedded.org/mt/89621292/3647476
> Group Owner: openembedded-core+owner@lists.openembedded.org
> Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub [kweihmann@outlook.com]
> -=-=-=-=-=-=-=-=-=-=-=-
> 

Re: [OE-core] [PATCH 1/4] pip_install_wheel: don't lazy assign PIPINSTALLARGS

[Ross Burton]

[-- Attachment #1: Type: text/plain, Size: 2227 bytes --]

Personally I can’t see any reason to extend the pip install arguments. What
option do you think would be used by a recipe?

Ross

On Mon, 7 Mar 2022 at 20:06, Konrad Weihmann <kweihmann@outlook.com> wrote:

>
> On 07.03.22 20:42, Ross Burton wrote:
> > If we expect users to extend this we should use =, as otherwise a recipe
> > that does += will replace the default value.
> >
> > Signed-off-by: Ross Burton <ross.burton@arm.com>
> > ---
> >   meta/classes/pip_install_wheel.bbclass | 2 +-
> >   1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/meta/classes/pip_install_wheel.bbclass
> b/meta/classes/pip_install_wheel.bbclass
> > index c1680a24ed..febcc8e445 100644
> > --- a/meta/classes/pip_install_wheel.bbclass
> > +++ b/meta/classes/pip_install_wheel.bbclass
> > @@ -10,7 +10,7 @@ PIP_INSTALL_PACKAGE ?=
> "${@guess_pip_install_package_name(d)}"
> >   PIP_INSTALL_DIST_PATH ?= "${@d.getVar('SETUPTOOLS_SETUP_PATH') or
> d.getVar('B')}/dist"
> >   PYPA_WHEEL ??=
> "${PIP_INSTALL_DIST_PATH}/${PIP_INSTALL_PACKAGE}-*-*.whl"
> >
> > -PIP_INSTALL_ARGS ?= "\
> > +PIP_INSTALL_ARGS = "\
> >       -vvvv \
> >       --ignore-installed \
> >       --no-cache \
> >
> >
>
> Hmm all of the classes use ?= and I think for a reason - this isn't the
> fault of the bbclass but of the recipe in the end, as += is known to
> cause this depending on where the additions have been added (before the
> inherit or after it) - I think the better way would be to use
> PIP_INSTALL_ARGS:append in the actual recipe - guess the fallout would
> be way less than the one I expect from this change.
>
> The big question is if that should changeable by the user or not - I
> personally think it should be
>
> >
> > -=-=-=-=-=-=-=-=-=-=-=-
> > Links: You receive all messages sent to this group.
> > View/Reply Online (#162860):
> https://lists.openembedded.org/g/openembedded-core/message/162860
> > Mute This Topic: https://lists.openembedded.org/mt/89621292/3647476
> > Group Owner: openembedded-core+owner@lists.openembedded.org
> > Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub [
> kweihmann@outlook.com]
> > -=-=-=-=-=-=-=-=-=-=-=-
> >
>

[-- Attachment #2: Type: text/html, Size: 3389 bytes --]

Re: [OE-core] [PATCH 1/4] pip_install_wheel: don't lazy assign PIPINSTALLARGS

[Konrad Weihmann]

On 07.03.22 22:17, Ross Burton wrote:
> Personally I can’t see any reason to extend the pip install arguments. 
> What option do you think would be used by a recipe?

For instance the --user option is something that I used once in a setup.
Plus --global-option for passing additional options into the 
installation scripting.
I give you that, that these are used rarely - and still could be used, 
if the recipes would be tweaked to append after the inherit statement.

And in my mind it was always like a variable with a weak assignment 
could be changed by the user, the ones with a hard assignment shouldn't 
be touched - but that could be just my mind 😅

> 
> Ross
> 
> On Mon, 7 Mar 2022 at 20:06, Konrad Weihmann <kweihmann@outlook.com 
> <mailto:kweihmann@outlook.com>> wrote:
> 
> 
>     On 07.03.22 20:42, Ross Burton wrote:
>      > If we expect users to extend this we should use =, as otherwise a
>     recipe
>      > that does += will replace the default value.
>      >
>      > Signed-off-by: Ross Burton <ross.burton@arm.com
>     <mailto:ross.burton@arm.com>>
>      > ---
>      >   meta/classes/pip_install_wheel.bbclass | 2 +-
>      >   1 file changed, 1 insertion(+), 1 deletion(-)
>      >
>      > diff --git a/meta/classes/pip_install_wheel.bbclass
>     b/meta/classes/pip_install_wheel.bbclass
>      > index c1680a24ed..febcc8e445 100644
>      > --- a/meta/classes/pip_install_wheel.bbclass
>      > +++ b/meta/classes/pip_install_wheel.bbclass
>      > @@ -10,7 +10,7 @@ PIP_INSTALL_PACKAGE ?=
>     "${@guess_pip_install_package_name(d)}"
>      >   PIP_INSTALL_DIST_PATH ?= "${@d.getVar('SETUPTOOLS_SETUP_PATH')
>     or d.getVar('B')}/dist"
>      >   PYPA_WHEEL ??=
>     "${PIP_INSTALL_DIST_PATH}/${PIP_INSTALL_PACKAGE}-*-*.whl"
>      >
>      > -PIP_INSTALL_ARGS ?= "\
>      > +PIP_INSTALL_ARGS = "\
>      >       -vvvv \
>      >       --ignore-installed \
>      >       --no-cache \
>      >
>      >
> 
>     Hmm all of the classes use ?= and I think for a reason - this isn't the
>     fault of the bbclass but of the recipe in the end, as += is known to
>     cause this depending on where the additions have been added (before the
>     inherit or after it) - I think the better way would be to use
>     PIP_INSTALL_ARGS:append in the actual recipe - guess the fallout would
>     be way less than the one I expect from this change.
> 
>     The big question is if that should changeable by the user or not - I
>     personally think it should be
> 
>      >
>      > -=-=-=-=-=-=-=-=-=-=-=-
>      > Links: You receive all messages sent to this group.
>      > View/Reply Online (#162860):
>     https://lists.openembedded.org/g/openembedded-core/message/162860
>     <https://lists.openembedded.org/g/openembedded-core/message/162860>
>      > Mute This Topic:
>     https://lists.openembedded.org/mt/89621292/3647476
>     <https://lists.openembedded.org/mt/89621292/3647476>
>      > Group Owner: openembedded-core+owner@lists.openembedded.org
>     <mailto:openembedded-core%2Bowner@lists.openembedded.org>
>      > Unsubscribe:
>     https://lists.openembedded.org/g/openembedded-core/unsub
>     <https://lists.openembedded.org/g/openembedded-core/unsub>
>     [kweihmann@outlook.com <mailto:kweihmann@outlook.com>]
>      > -=-=-=-=-=-=-=-=-=-=-=-
>      >
> 

Re: [OE-core] [PATCH 1/4] pip_install_wheel: don't lazy assign PIPINSTALLARGS

[Ross Burton]

On Tue, 8 Mar 2022 at 07:40, Konrad Weihmann <kweihmann@outlook.com> wrote:
> For instance the --user option is something that I used once in a setup.

This is the pip_install_wheel class for installing wheels to the
system library directory, so --user wouldn't be useful surely?

Ross