| Message ID | 20240415131443.7295-1-marex@denx.de |
|---|---|
| State | New |
| Headers | show |
| Series | [meta-python2] layer.conf: update LAYERSERIES_COMPAT for scarthgap | expand |
Hi Marek, Last commit on this layer was done 2 years ago and it was at the time to add Kirkstone compatibility, nothing more since then. Python 2 is dead since 1/1/2020. That means that we (python dev) will not improve it anymore after that day, even if someone finds a security problem in it. You should upgrade to Python 3 as soon as you can. https://www.python.org/doc/sunset-python-2 In my opinion the OE community should no longer promote its use. Jose Marek Vasut via lists.openembedded.org <marex=denx.de@lists.openembedded.org> escreveu (segunda, 15/04/2024 à(s) 14:15): > Support both kirkstone and scarthgap as that seems to work well. > > Signed-off-by: Marek Vasut <marex@denx.de> > --- > conf/layer.conf | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/conf/layer.conf b/conf/layer.conf > index 38ef579..0f228ca 100644 > --- a/conf/layer.conf > +++ b/conf/layer.conf > @@ -14,7 +14,7 @@ LAYERVERSION_meta-python2 = "1" > > LAYERDEPENDS_meta-python2 = "core openembedded-layer" > > -LAYERSERIES_COMPAT_meta-python2 = "kirkstone" > +LAYERSERIES_COMPAT_meta-python2 = "kirkstone scarthgap" > > LICENSE_PATH += "${LAYERDIR}/licenses" > > -- > 2.43.0 > > > -=-=-=-=-=-=-=-=-=-=-=- > Links: You receive all messages sent to this group. > View/Reply Online (#109961): > https://lists.openembedded.org/g/openembedded-devel/message/109961 > Mute This Topic: https://lists.openembedded.org/mt/105535514/5052612 > Group Owner: openembedded-devel+owner@lists.openembedded.org > Unsubscribe: https://lists.openembedded.org/g/openembedded-devel/unsub [ > quaresma.jose@gmail.com] > -=-=-=-=-=-=-=-=-=-=-=- > >
On 4/15/24 3:49 PM, Jose Quaresma wrote: > Hi Marek, Hi, > Last commit on this layer was done 2 years ago and it was at the time to > add Kirkstone compatibility, nothing more since then. > Python 2 is dead since 1/1/2020. That means that we (python dev) will not > improve it anymore after that day, > even if someone finds a security problem in it. You should upgrade to > Python 3 as soon as you can. > > https://www.python.org/doc/sunset-python-2 > > In my opinion the OE community should no longer promote its use. I am aware of this, for me this is just a band-aid to keep meta-qt5 building for a while longer. I can work around this in bblayers.conf or local.conf too, I would like to avoid that however.
This is a hard no from me. On Mon, Apr 15, 2024 at 6:59 AM Marek Vasut via lists.openembedded.org <marex=denx.de@lists.openembedded.org> wrote: > On 4/15/24 3:49 PM, Jose Quaresma wrote: > > Hi Marek, > > Hi, > > > Last commit on this layer was done 2 years ago and it was at the time to > > add Kirkstone compatibility, nothing more since then. > > Python 2 is dead since 1/1/2020. That means that we (python dev) will not > > improve it anymore after that day, > > even if someone finds a security problem in it. You should upgrade to > > Python 3 as soon as you can. > > > > https://www.python.org/doc/sunset-python-2 > > > > In my opinion the OE community should no longer promote its use. > > I am aware of this, for me this is just a band-aid to keep meta-qt5 > building for a while longer. I can work around this in bblayers.conf or > local.conf too, I would like to avoid that however. > > -=-=-=-=-=-=-=-=-=-=-=- > Links: You receive all messages sent to this group. > View/Reply Online (#109963): > https://lists.openembedded.org/g/openembedded-devel/message/109963 > Mute This Topic: https://lists.openembedded.org/mt/105535514/924729 > Group Owner: openembedded-devel+owner@lists.openembedded.org > Unsubscribe: https://lists.openembedded.org/g/openembedded-devel/unsub [ > ticotimo@gmail.com] > -=-=-=-=-=-=-=-=-=-=-=- > >
Yes to Tim's hard no. Qt5 itself is EOL for almost a year now (don't tell me about their commercial offers, ok?). I would put something similar to meta-gpl2 into that layer instead actually: https://git.yoctoproject.org/meta-gplv2/commit/?id=43bf0e8d5985945d19d01f94bfbbda420c4435f3 Alex On Mon, 15 Apr 2024 at 17:21, Tim Orling via lists.openembedded.org <ticotimo=gmail.com@lists.openembedded.org> wrote: > > This is a hard no from me. > > On Mon, Apr 15, 2024 at 6:59 AM Marek Vasut via lists.openembedded.org <marex=denx.de@lists.openembedded.org> wrote: >> >> On 4/15/24 3:49 PM, Jose Quaresma wrote: >> > Hi Marek, >> >> Hi, >> >> > Last commit on this layer was done 2 years ago and it was at the time to >> > add Kirkstone compatibility, nothing more since then. >> > Python 2 is dead since 1/1/2020. That means that we (python dev) will not >> > improve it anymore after that day, >> > even if someone finds a security problem in it. You should upgrade to >> > Python 3 as soon as you can. >> > >> > https://www.python.org/doc/sunset-python-2 >> > >> > In my opinion the OE community should no longer promote its use. >> >> I am aware of this, for me this is just a band-aid to keep meta-qt5 >> building for a while longer. I can work around this in bblayers.conf or >> local.conf too, I would like to avoid that however. >> >> >> > > -=-=-=-=-=-=-=-=-=-=-=- > Links: You receive all messages sent to this group. > View/Reply Online (#109964): https://lists.openembedded.org/g/openembedded-devel/message/109964 > Mute This Topic: https://lists.openembedded.org/mt/105535514/1686489 > Group Owner: openembedded-devel+owner@lists.openembedded.org > Unsubscribe: https://lists.openembedded.org/g/openembedded-devel/unsub [alex.kanavin@gmail.com] > -=-=-=-=-=-=-=-=-=-=-=- >
If you need it just for some experiments than feel free to use: https://github.com/shr-project/meta-python2/commits/jansa/master/ which I've already updated (for built time benchmark which historically included qtwebengine as well https://github.com/shr-project/test-oe-build-time). But for anything more serious work with your customer to switch to something more supported, there are still minor upgrades to meta-qt5 (last in https://github.com/meta-qt5/meta-qt5/pull/556), but it's a dead end and switch to something else should have higher priority than switch to scarthgap (e.g. switching to meta-qt6). Cheers, On Mon, Apr 15, 2024 at 3:15 PM Marek Vasut <marex@denx.de> wrote: > > Support both kirkstone and scarthgap as that seems to work well. > > Signed-off-by: Marek Vasut <marex@denx.de> > --- > conf/layer.conf | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/conf/layer.conf b/conf/layer.conf > index 38ef579..0f228ca 100644 > --- a/conf/layer.conf > +++ b/conf/layer.conf > @@ -14,7 +14,7 @@ LAYERVERSION_meta-python2 = "1" > > LAYERDEPENDS_meta-python2 = "core openembedded-layer" > > -LAYERSERIES_COMPAT_meta-python2 = "kirkstone" > +LAYERSERIES_COMPAT_meta-python2 = "kirkstone scarthgap" > > LICENSE_PATH += "${LAYERDIR}/licenses" > > -- > 2.43.0 >
On 4/15/24 7:48 PM, Martin Jansa wrote: > If you need it just for some experiments than feel free to use: > https://github.com/shr-project/meta-python2/commits/jansa/master/ > which I've already updated (for built time benchmark which > historically included qtwebengine as well > https://github.com/shr-project/test-oe-build-time). > > But for anything more serious work with your customer to switch to > something more supported, there are still minor upgrades to meta-qt5 > (last in https://github.com/meta-qt5/meta-qt5/pull/556), but it's a > dead end and switch to something else should have higher priority than > switch to scarthgap (e.g. switching to meta-qt6). I am aware of qt5 being a dead end, but I also think if making this stuff build with scarthgap is a one-liner change, I don't really see why not enable people to build it. The security implications have to be considered by the users. Maybe this should also be spelled out in the README, for both the meta-python2 and meta-qt5. I can send such a patch ?
On Tue, 16 Apr 2024 at 15:55, Marek Vasut via lists.openembedded.org <marex=denx.de@lists.openembedded.org> wrote: > I am aware of qt5 being a dead end, but I also think if making this > stuff build with scarthgap is a one-liner change, I don't really see why > not enable people to build it. > > The security implications have to be considered by the users. Maybe this > should also be spelled out in the README, for both the meta-python2 and > meta-qt5. I can send such a patch ? Again, no. We need to draw the line somewhere, and this is where we draw it. Martin can of course still push the patch, but I'd be disappointed if he did. Alex
-----Original Message----- From: openembedded-devel@lists.openembedded.org <openembedded-devel@lists.openembedded.org> On Behalf Of Marek Vasut via lists.openembedded.org Sent: Tuesday, April 16, 2024 15:55 To: Martin Jansa <martin.jansa@gmail.com> Cc: openembedded-devel@lists.openembedded.org Subject: Re: [oe] [meta-python2][PATCH] layer.conf: update LAYERSERIES_COMPAT for scarthgap > On 4/15/24 7:48 PM, Martin Jansa wrote: > > If you need it just for some experiments than feel free to use: > > https://github.com/shr-project/meta-python2/commits/jansa/master/ > > which I've already updated (for built time benchmark which > > historically included qtwebengine as well > > https://github.com/shr-project/test-oe-build-time). > > > > But for anything more serious work with your customer to switch to > > something more supported, there are still minor upgrades to meta-qt5 > > (last in https://github.com/meta-qt5/meta-qt5/pull/556), but it's a > > dead end and switch to something else should have higher priority than > > switch to scarthgap (e.g. switching to meta-qt6). > > I am aware of qt5 being a dead end, but I also think if making this stuff build with scarthgap is a one-liner change, I don't really see why not enable people to build it. > > The security implications have to be considered by the users. Maybe this should also be spelled out in the README, for both the meta-python2 and meta-qt5. I can send such a patch ? Following works for yocto kirkstone + meta-lts-mixin kirkstone/go + meta-lts-mixin kirkstone/rust + meta-virtualization master LAYERSERIES_COMPAT_virtualization-layer += "kirkstone" meta-qt5 should be able to do something similar with meta-python2 if meta-python2 is frozen...
On Tue, 16 Apr 2024 at 16:19, Peter Marko via lists.openembedded.org <peter.marko=siemens.com@lists.openembedded.org> wrote: > Following works for yocto kirkstone + meta-lts-mixin kirkstone/go + meta-lts-mixin kirkstone/rust + meta-virtualization master > LAYERSERIES_COMPAT_virtualization-layer += "kirkstone" > > meta-qt5 should be able to do something similar with meta-python2 if meta-python2 is frozen... If I remember right, meta-qt5 does not need target python2, it need only native python2 because the qt5 web widget contains a badly outdated chromium. So meta-qt5 could just copy the python2 recipe directly into the layer perhaps and drop meta-python2 dependency. Alex
Target python recipes are already blacklisted since
https://github.com/shr-project/meta-python2/commit/10c65fc18998d302f34634ffceaf31ab335056dd
so meta-python2 already provides just python-native and necessary
bbclasses (as pythonnative.bbclass and few others).
So I don't want to add another identical python2 recipe in meta-qt5
(which will be included for all meta-qt5 users) while now only a few
recipes in meta-qt5 need meta-python2 layer to be included:
./recipes-qt/qt-kiosk-browser/qt-kiosk-browser_git.bb: if
'meta-python2' not in d.getVar('BBFILE_COLLECTIONS').split():
./recipes-qt/qt-kiosk-browser/qt-kiosk-browser_git.bb: raise
bb.parse.SkipRecipe('qtwebengine dependency requires meta-python2 to
be present.')
./recipes-qt/qt5/qtpdf_git.bb:# still depends on python2
./recipes-qt/qt5/qtpdf_git.bb:# First patch skips "python2" dependency
checks for the pdf module
./recipes-qt/qt5/qtpdf_git.bb:
file://0001-configure.json-remove-python2-dependency.patch \
./recipes-qt/qt5/qtpdf_git.bb:PACKAGECONFIG[webengine-pdf] =
"-feature-webengine-python2,-no-feature-webengine-python2,python3-native"
./recipes-qt/qt5/qtpdf/0001-configure.json-remove-python2-dependency.patch:Subject:
[PATCH] configure.json: remove python2 dependency
./recipes-qt/qt5/qtpdf/0001-configure.json-remove-python2-dependency.patch:-
&& features.webengine-python2
./recipes-qt/qt5/qtwebengine_git.bb:inherit
${@bb.utils.contains("BBFILE_COLLECTIONS", "meta-python2",
"pythonnative", "", d)}
./recipes-qt/qt5/qtwebengine_git.bb: if 'meta-python2' not in
d.getVar('BBFILE_COLLECTIONS').split():
./recipes-qt/qt5/qtwebengine_git.bb: raise
bb.parse.SkipRecipe('Requires meta-python2 to be present.')
./recipes-qt/qt5/qt5-creator_git.bb: if 'meta-python2' not in
d.getVar('BBFILE_COLLECTIONS').split():
./recipes-qt/qt5/qt5-creator_git.bb: raise
bb.parse.SkipRecipe('Requires meta-python2 to be present.')
./recipes-qt/qt5/qtwebview_git.bb: if 'meta-python2' not in
d.getVar('BBFILE_COLLECTIONS').split():
./recipes-qt/qt5/qtwebview_git.bb: raise
bb.parse.SkipRecipe('Requires meta-python2 to be present.')
That's why I've included scarthgap compatibility in my fork of meta-python2.
As there is strong opinion against declaring scarthgap compatibility
in official meta-python2 from Timo, Alex and Jose, I don't plan to
merge Marek's change (nor my version with additional fixes for
Upstream-Status etc).
Maybe usable compromise would be to delete all (already blacklisted)
target recipes from existing meta-python2 and leave only the
python-native to make it even more clear what this layer is
temporarily used for (maybe even with another blacklist by default
like I_REALLY_SWEAR_TO_MIGRATE_TO_PYTHON3_OR_DIE_TRYING) which IMHO
makes better sense than importing this old unsupported python-native
to meta-qt5.
Cheers,
On Tue, Apr 16, 2024 at 4:27 PM Alexander Kanavin
<alex.kanavin@gmail.com> wrote:
>
> On Tue, 16 Apr 2024 at 16:19, Peter Marko via lists.openembedded.org
> <peter.marko=siemens.com@lists.openembedded.org> wrote:
> > Following works for yocto kirkstone + meta-lts-mixin kirkstone/go + meta-lts-mixin kirkstone/rust + meta-virtualization master
> > LAYERSERIES_COMPAT_virtualization-layer += "kirkstone"
> >
> > meta-qt5 should be able to do something similar with meta-python2 if meta-python2 is frozen...
>
> If I remember right, meta-qt5 does not need target python2, it need
> only native python2 because the qt5 web widget contains a badly
> outdated chromium. So meta-qt5 could just copy the python2 recipe
> directly into the layer perhaps and drop meta-python2 dependency.
>
> Alex
On Tue, 16 Apr 2024 at 17:03, Martin Jansa <martin.jansa@gmail.com> wrote: > As there is strong opinion against declaring scarthgap compatibility > in official meta-python2 from Timo, Alex and Jose, I don't plan to > merge Marek's change (nor my version with additional fixes for > Upstream-Status etc). There's more to it. It's a protest against people focusing on obsolete retro computing stuff because of their poor product planning, while offering no help to keep things current. It took me two months to update python to 3.12 while ensuring nothing breaks in core or meta-oe. Something you Marek perhaps don't need right now, but will certainly need in the future. Did you help? Alex
On Wed, Apr 17, 2024 at 2:00 AM Alexander Kanavin via lists.openembedded.org <alex.kanavin=gmail.com@lists.openembedded.org> wrote: > On Tue, 16 Apr 2024 at 17:03, Martin Jansa <martin.jansa@gmail.com> wrote: > > As there is strong opinion against declaring scarthgap compatibility > > in official meta-python2 from Timo, Alex and Jose, I don't plan to > > merge Marek's change (nor my version with additional fixes for > > Upstream-Status etc). > > There's more to it. It's a protest against people focusing on obsolete > retro computing stuff because of their poor product planning, while > offering no help to keep things current. It took me two months to > update python to 3.12 while ensuring nothing breaks in core or > meta-oe. Something you Marek perhaps don't need right now, but will > certainly need in the future. Did you help? > To echo what Alex is saying here (I have spent months of my life, uncompensated for the effort, to bring the pyproject.toml and other new PEP-517 build backend classes into oe-core, with a lot of very hard effort from RP and Ross). In addition, with my OpenEmbedded Board member hat on and my OpenEmbedded TSC member hat on, I take this stance: We as an organization in no way whatsoever want anyone to think it is ok to continue to use python2. You have had well over a decade of warning and foreshadowing. The licensing probably means we do not have any actual legal liability exposure, but in our modern litigious society, this is no guarantee. You have had two years of kirkstone support and still have two more years before kirkstone is EOL. This is the end of the line. This is where this stops. Plan your way out of this situation and move on. Tim We cannot control what users and companies want to bring into their own layers. They are welcome to at their > > Alex > > -=-=-=-=-=-=-=-=-=-=-=- > Links: You receive all messages sent to this group. > View/Reply Online (#110039): > https://lists.openembedded.org/g/openembedded-devel/message/110039 > Mute This Topic: https://lists.openembedded.org/mt/105535514/924729 > Group Owner: openembedded-devel+owner@lists.openembedded.org > Unsubscribe: https://lists.openembedded.org/g/openembedded-devel/unsub [ > ticotimo@gmail.com] > -=-=-=-=-=-=-=-=-=-=-=- > >
diff --git a/conf/layer.conf b/conf/layer.conf index 38ef579..0f228ca 100644 --- a/conf/layer.conf +++ b/conf/layer.conf @@ -14,7 +14,7 @@ LAYERVERSION_meta-python2 = "1" LAYERDEPENDS_meta-python2 = "core openembedded-layer" -LAYERSERIES_COMPAT_meta-python2 = "kirkstone" +LAYERSERIES_COMPAT_meta-python2 = "kirkstone scarthgap" LICENSE_PATH += "${LAYERDIR}/licenses"
Support both kirkstone and scarthgap as that seems to work well. Signed-off-by: Marek Vasut <marex@denx.de> --- conf/layer.conf | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)