recipeutils: add optional stable_upgrade parameter to get_recipe_upgrade_status
PATCH v3 1/6 recipeutils: add optional stable_upgrade parameter to get_recipe_upgrade_status
From: Chen Qi <[email protected]>
We want the ability to do stable version upgrades for recipes. To this end, add an optional stable_upgrade parameter to the get_recipe_upgrade_status function, which defaults to False and when enabled will try to get the latest stable version of the recipe.
The UPSTREAM_STABLE_RELEASE_REGEX is respected. If a recipe sets it, it will be used as the filter_regex. If it's not set explicitly, it means that there's no stable updates or the recipe hasn't been checked yet.
--- meta/lib/oe/recipeutils.py | 23 +++++++++++++++++------ 1 file changed, 17 insertions(+), 6 deletions(-)
meta/lib/oe/recipeutils.py +17 -6
@@ -1009,7 +1009,7 @@ def get_recipe_pv_with_pfx_sfx(pv, uri_type): return (pv, pfx, sfx) -def get_recipe_upstream_version(rd): +def get_recipe_upstream_version(rd, stable_upgrade): """ Get upstream version of recipe using bb.fetch2 methods with support for http, https, ftp and git.
@@ -1080,7 +1080,15 @@ def get_recipe_upstream_version(rd): except bb.fetch2.FetchError as e: bb.warn("Unable to obtain latest revision: {}".format(e)) else: - pupver = ud.method.latest_versionstring(ud, rd) + if stable_upgrade: + stable_release_regex = rd.getVar("UPSTREAM_STABLE_RELEASE_REGEX") + if stable_release_regex: + pupver = ud.method.latest_versionstring(ud, rd, filter_regex=stable_release_regex) + else: + # Not explicitly setting "UPSTREAM_STABLE_RELEASE_REGEX" means there's no stable upgrade + pupver = (ru['current_version'], None) + else: + pupver = ud.method.latest_versionstring(ud, rd) (upversion, revision) = pupver if upversion:
@@ -1094,8 +1102,8 @@ def get_recipe_upstream_version(rd): return ru -def _get_recipe_upgrade_status(data): - uv = get_recipe_upstream_version(data) +def _get_recipe_upgrade_status(data, stable_upgrade): + uv = get_recipe_upstream_version(data, stable_upgrade) pn = data.getVar('PN') cur_ver = uv['current_version']
@@ -1119,9 +1127,10 @@ def _get_recipe_upgrade_status(data): return {'pn':pn, 'status':status, 'cur_ver':cur_ver, 'next_ver':next_ver, 'maintainer':maintainer, 'revision':revision, 'no_upgrade_reason':no_upgrade_reason} -def get_recipe_upgrade_status(recipes=None): +def get_recipe_upgrade_status(recipes=None, stable_upgrade=False): pkgs_list = [] data_copy_list = [] + stable_copy_list = [] copy_vars = ('SRC_URI', 'PV', 'DL_DIR',
@@ -1134,6 +1143,7 @@ def get_recipe_upgrade_status(recipes=None): 'UPSTREAM_CHECK_REGEX', 'UPSTREAM_CHECK_URI', 'UPSTREAM_VERSION_UNKNOWN', + 'UPSTREAM_STABLE_RELEASE_REGEX', 'RECIPE_MAINTAINER', 'RECIPE_NO_UPDATE_REASON', 'RECIPE_UPSTREAM_VERSION',
@@ -1180,12 +1190,13 @@ def get_recipe_upgrade_status(recipes=None): data_copy.setVar(k, data.getVar(k)) data_copy_list.append(data_copy) + stable_copy_list.append(stable_upgrade) recipeincludes[data.getVar('FILE')] = {'bbincluded':data.getVar('BBINCLUDED').split(),'pn':data.getVar('PN')} from concurrent.futures import ProcessPoolExecutor with ProcessPoolExecutor(max_workers=utils.cpu_count()) as executor: - pkgs_list = executor.map(_get_recipe_upgrade_status, data_copy_list) + pkgs_list = executor.map(_get_recipe_upgrade_status, data_copy_list, stable_copy_list) return _group_recipes(pkgs_list, _get_common_include_recipes(recipeincludes))
Signature · 1 line
2.34.1
PATCH v3 2/6 upstream-stable-release-point.bbclass: add bbclass for stable point upgrade
From: Chen Qi <[email protected]>
When inheriting this bbclass, the recipe should have been checked to be sure that the last part bump (e.g., 1.0.0 -> 1.0.1 or 34 -> 34.1) is a stable version upgrade.
--- .../upstream-stable-release-point.bbclass | 22 +++++++++++++++++++ 1 file changed, 22 insertions(+) create mode 100644 meta/classes-recipe/upstream-stable-release-point.bbclass
meta/classes-recipe/upstream-stable-release-point.bbclass new file +22 -0
@@ -0,0 +1,22 @@ +# +# Copyright OpenEmbedded Contributors +# +# SPDX-License-Identifier: MIT +# + +# +# This bbclass is expected to be inherited by recipes explicitly. +# If a recipe's version is separated by point and its last part's +# bump is a stable upgrade, then it can inherit this bbclass. +# + +def get_majmin_version_regex(d): + pv = d.getVar('PV') + v_parts = pv.split('.') + if len(v_parts) == 1: + return pv + else: + return '\.'.join(v_parts[:-1]) + +MAJMIN_VERSION_REGEX = "${@get_majmin_version_regex(d)}" +UPSTREAM_STABLE_RELEASE_REGEX = "${MAJMIN_VERSION_REGEX}\.\d"
Signature · 1 line
2.34.1
Re: [OE-core][PATCH V3 2/6] upstream-stable-release-point.bbclass: add bbclass for stable point upgrade
meta/classes-recipe/upstream-stable-release-point.bbclass line 19
Quoted 7 lines · On Sat, 9 May 2026 at 07:30, <[email protected]> wrote:
meta/classes-recipe/upstream-stable-release-point.bbclass new file +7
-0+def get_majmin_version_regex(d): + pv = d.getVar('PV') + v_parts = pv.split('.') + if len(v_parts) == 1: + return pv + else: + return '\.'.join(v_parts[:-1])
To be honest, I would not special-case the len(v_parts) == 1, as the class becomes somewhat inconsistent:
x -> x.1 is supported
x.y -> x.y.1 is not supported (instead it treats the stable versions as x.y -> x.y+1 )
These are two different schemes, so I would not mix them in a single class, and support only:
x.y -> x.y+1
x.y.z -> x.y.z+1
x.y.z.zz -> x.y.z.zz+1
etc
For x->x.1 we can make a different class,
'upstream-stable-release-append-number' maybe ?
meta/classes-recipe/upstream-stable-release-point.bbclass line 22
Quoted 2 lines
meta/classes-recipe/upstream-stable-release-point.bbclass new file +2
-0+MAJMIN_VERSION_REGEX = "${@get_majmin_version_regex(d)}" +UPSTREAM_STABLE_RELEASE_REGEX = "${MAJMIN_VERSION_REGEX}\.\d"
\.\d should be (\.\d+)+ to allow numbers bigger than 9 and multiple .number parts one after another.
Alex
Re: [OE-core][PATCH V3 2/6] upstream-stable-release-point.bbclass: add bbclass for stable point upgrade
Quoted 10 lines · Alexander Kanavin
To be honest, I would not special-case the
len(v_parts)== 1, as the class becomes somewhat inconsistent:x -> x.1 is supported
x.y -> x.y.1 is not supported (instead it treats the stable versions as x.y -> x.y+1 )These are two different schemes, so I would not mix them in a single class, and support only:
Quoted 7 lines
x.y -> x.y+1
x.y.z -> x.y.z+1
x.y.z.zz -> x.y.z.zz+1
etcFor x->x.1 we can make a different class,
'upstream-stable-release-append-number' maybe ?
On further thought, the class should support only
major.minor.point[.subpoint] schemes.
The problem with removing only the last number is when upstream issues:
x.y.z -> x.y.z.zz, we update to that, and then there's x.y.z.zz -> x.y.z+1 - this would not be considered a stable update, even though it is.
It's a bit complicated to get it right :)
So basically:
def get_majmin_version_regex(d):
pv = d.getVar('PV')
v_parts_majmin = pv.split('.')[:2]
return '\.'.join(v_parts_majmin)...UPSTREAM_STABLE_RELEASE_REGEX = "${MAJMIN_VERSION_REGEX}(\.\d+)+"
Alex
Re: [OE-core][PATCH V3 2/6] upstream-stable-release-point.bbclass: add bbclass for stable point upgrade
In reply to Alexander KanavinQuoted 10 lines · Alexander Kanavin
meta/classes-recipe/upstream-stable-release-point.bbclass line 19
Quoted 7 lines · On Sat, 9 May 2026 at 07:30, <[email protected]> wrote:
meta/classes-recipe/upstream-stable-release-point.bbclass new file +7
-0+def get_majmin_version_regex(d): + pv = d.getVar('PV') + v_parts = pv.split('.') + if len(v_parts) == 1: + return pv + else: + return '\.'.join(v_parts[:-1])To be honest, I would not special-case the
len(v_parts)== 1, as the class becomes somewhat inconsistent:
Hi Alex,
The check is to avoid the function returning empty string. At that time, I was dealing with systemd and kmod.
I agree that this x -> x.1 special case introduces inconsistency.
Quoted 8 lines
x -> x.1 is supported
x.y -> x.y.1 is not supported (instead it treats the stable versions as x.y -> x.y+1 )These are two different schemes, so I would not mix them in a single class, and support only:
Quoted 7 lines
x.y -> x.y+1
x.y.z -> x.y.z+1
x.y.z.zz -> x.y.z.zz+1
etcFor x->x.1 we can make a different class,
'upstream-stable-release-append-number' maybe ?
If we do so, recipes like systemd/kmod cannot inherit this bbclass as it will do uprevs like 259.1 -> 259.2 as well as 259 -> 259.1.
I mean, they can only inherit one such bbclass.
Ideally, I'd like things to be in one bbclass, maybe with some var controlling the behavior.
Quoted 5 lines
meta/classes-recipe/upstream-stable-release-point.bbclass line 22
Quoted 3 lines
meta/classes-recipe/upstream-stable-release-point.bbclass new file +2
-0+MAJMIN_VERSION_REGEX = "${@get_majmin_version_regex(d)}" +UPSTREAM_STABLE_RELEASE_REGEX = "${MAJMIN_VERSION_REGEX}\.\d"\.\d should be (\.\d+)+ to allow numbers bigger than 9 and multiple .number parts one after another.
You're right about this '+'.
With your review, I'm now thinking about using something like below:
Introduce one configurable variable for this bbclass: MAJMIN_VERSION_PARTS ?= ""
By default, it's not set and the majmin version should be the [:-1] part (I mean for 1.2.3, majmin is 1.2). And for recipes like systemd/kmod, when inheriting this bbclass, they should also set MAJMIN_VERSION_PARTS = "1", so the majmin version is [:1]. Recipes who want to do x.y -> x.y.1 or x.y -> x.y.1.1 can set MAJMIN_VERSION_PARTS = "2", so that the majmin version is [:2] which is x.y in this case.
The number of recipes setting this var should be few.
What do you think about the above solution?
Regards,
Qi
Quoted 2 lines
Alex
Re: [OE-core][PATCH V3 2/6] upstream-stable-release-point.bbclass: add bbclass for stable point upgrade
Quoted 10 lines · ChenQi
Introduce one configurable variable for this bbclass:
MAJMIN_VERSION_PARTS?= ""By default, it's not set and the majmin version should be the [:-1] part (I mean for 1.2.3, majmin is 1.2). And for recipes like systemd/kmod, when inheriting this bbclass, they should also set
MAJMIN_VERSION_PARTS= "1", so the majmin version is [:1]. Recipes who want to do x.y -> x.y.1 or x.y -> x.y.1.1 can setMAJMIN_VERSION_PARTS= "2", so that the majmin version is [:2] which is x.y in this case.
Quoted 3 lines
The number of recipes setting this var should be few.
What do you think about the above solution?
I agree with introducing the variable for configuration, but the problem with using the [:-1] default is that it sets up an upgrade 'trap' where this sequence in stable x.y releases:
x.y.z -> x.y.z.zz
x.y.z.zz -> x.y.z+1
will cause recipes to get stuck at x.y.z.zz.
My proposal is:
- let's drop the majmin naming, it's only suitable for major.minor schemes. I'd suggest naming it STABLE_VERSION_PARTS.
- then the class would slice that many parts from the front, and then add a regex that says 'anything after this is a maintenance release'.
STABLE_VERSION_PARTS ?= "2"
def get_stable_version_regex(d):
pv = d.getVar('PV')
stable_parts = pv.split('.')[:int(d.getVar('STABLE_VERSION_PARTS'))]
return '\.'.join(stable_parts)STABLE_VERSION_REGEX = "${@get_stable_version_regex(d)}" UPSTREAM_STABLE_RELEASE_REGEX = "${STABLE_VERSION_REGEX}.+"
I think this should cover everything common?
Alex
PATCH v3 3/6 vte: inherit upstream-stable-release-point
In reply to [email protected]From: Chen Qi <[email protected]>
The last version part bump for vte is a stable upgrade.
* Evidence
Its repo[1] has remote branches like:
origin/vte-0-76
origin/vte-0-78
origin/vte-0-80
origin/vte-0-82
origin/vte-0-84So for example, 0.82.2 -> 0.82.3 is a stable upgrade.
[1] https://gitlab.gnome.org/GNOME/vte.git
--- meta/recipes-support/vte/vte_0.82.2.bb | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
meta/recipes-support/vte/vte_0.82.2.bb +1 -1
@@ -15,7 +15,7 @@ DEPENDS = "fastfloat fmt glib-2.0 glib-2.0-native gtk+3 libpcre2 libxml2-native GIR_MESON_OPTION = 'gir' GIDOCGEN_MESON_OPTION = "docs" -inherit gnomebase gi-docgen features_check upstream-version-is-even gobject-introspection systemd vala +inherit gnomebase gi-docgen features_check upstream-version-is-even gobject-introspection systemd vala upstream-stable-release-point SRC_URI += "file://0001-Add-W_EXITCODE-macro-for-non-glibc-systems.patch \ file://0001-support-reproducibility-for-debug-sources.patch;patchdir=./subprojects/simdutf \
Signature · 1 line
2.34.1
PATCH v3 4/6 systemd: inherit upstream-stable-release-point
In reply to [email protected]From: Chen Qi <[email protected]>
The last version part bump for vte is a stable upgrade.
* Evidence
Its repo[1] has remote branches like:
origin/v256-stable
origin/v257-stable
origin/v258-stable
origin/v259-stable
origin/v260-stable[1] https://github.com/systemd/systemd.git
--- meta/recipes-core/systemd/systemd.inc | 2 ++ 1 file changed, 2 insertions(+)
meta/recipes-core/systemd/systemd.inc +2 -0
@@ -19,6 +19,8 @@ SRCREV = "b3d8fc43e9cb531d958c17ef2cd93b374bc14e8a" SRCBRANCH = "v259-stable" SRC_URI = "git://github.com/systemd/systemd.git;protocol=https;branch=${SRCBRANCH};tag=v${PV}" +inherit upstream-stable-release-point + CVE_PRODUCT = "systemd" CVE_STATUS[CVE-2019-3815] = "not-applicable-platform: only applied to RHEL"
Signature · 1 line
2.34.1
PATCH v3 5/6 kmod: inherit upstream-stable-release-point
In reply to [email protected]From: Chen Qi <[email protected]>
The last version part bump for vte is a stable upgrade.
* Evidence
Its repo contains a branch called:
origin/kmod-34.y
Its tags are like:
v32
v33
v34
v34.1
So we can infer that the maintenance policy for kmod is that upgrades such as 33 -> 34 contains features and upgrades such as 34 -> 34.1 is a stable upgrade. And such maintenance policy is likely to continue.
[1] https://git.kernel.org/pub/scm/utils/kernel/kmod/kmod.git
--- meta/recipes-kernel/kmod/kmod_34.2.bb | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
meta/recipes-kernel/kmod/kmod_34.2.bb +1 -1
@@ -13,7 +13,7 @@ LIC_FILES_CHKSUM = "file://COPYING;md5=a6f89e2100d9b6cdffcea4f398e37343 \ file://libkmod/COPYING;md5=a6f89e2100d9b6cdffcea4f398e37343 \ file://tools/COPYING;md5=751419260aa954499f7abaabaa882bbe \ " -inherit bash-completion gtk-doc manpages meson pkgconfig update-alternatives +inherit bash-completion gtk-doc manpages meson pkgconfig update-alternatives upstream-stable-release-point SRC_URI = "https://www.kernel.org/pub/linux/utils/kernel/${BPN}/${BP}.tar.xz \ file://depmod-search.conf \
Signature · 1 line
2.34.1
PATCH v3 6/6 util-linux: inherit upstream-stable-release-point
In reply to [email protected]From: Chen Qi <[email protected]>
The last version part bump for vte is a stable upgrade.
* Evidence
Its repo[1] contains branches like:
origin/stable/v2.39
origin/stable/v2.40
origin/stable/v2.41[1] https://github.com/util-linux/util-linux.git
--- meta/recipes-core/util-linux/util-linux.inc | 2 ++ 1 file changed, 2 insertions(+)
meta/recipes-core/util-linux/util-linux.inc +2 -0
@@ -25,4 +25,6 @@ SRC_URI = "${KERNELORG_MIRROR}/linux/utils/util-linux/v${MAJOR_VERSION}/util-lin SRC_URI[sha256sum] = "3330d873f0fceb5560b89a7dc14e4f3288bbd880e96903ed9b50ec2b5799e58b" +inherit upstream-stable-release-point + CVE_PRODUCT = "util-linux"
Signature · 1 line
2.34.1