| Message ID | 20250403-release-note-5-2-updates-fixes-v1-2-13d16361f17c@bootlin.com |
|---|---|
| State | Accepted |
| Headers | show |
| Series | Improvements and fixes for the 5.2 release | expand |
Hi Antonin, On 4/3/25 4:08 PM, Antonin Godard wrote: > Address some of the reviews received on the release notes and migration > guide for 5.2: > > - Remove the wget fetcher addition, I misread commit 0a9f90ff658e > ("tests/fetch: support setting PV in the wget fetcher") from BitBake, > and it actually affects the tests for the wget fetcher, not the > fetcher itself. > - Add a way to transition from the branch=nameX,nameY syntax. From what > I have tested, I achieve the same behavior when using one or the other > syntax. > - Remove the default value for SPDX_PACKAGE_VERSION in the release note. > - Fix a couple of typos. > > Suggested-by: Quentin Schulz <quentin.schulz@cherry.de> > Signed-off-by: Antonin Godard <antonin.godard@bootlin.com> > --- > documentation/migration-guides/migration-5.2.rst | 10 +++++++++- > documentation/migration-guides/release-notes-5.2.rst | 7 ++----- > 2 files changed, 11 insertions(+), 6 deletions(-) > > diff --git a/documentation/migration-guides/migration-5.2.rst b/documentation/migration-guides/migration-5.2.rst > index c54f054c7..59f21ff4c 100644 > --- a/documentation/migration-guides/migration-5.2.rst > +++ b/documentation/migration-guides/migration-5.2.rst > @@ -193,6 +193,14 @@ anymore:: > This was rarely used in the core repositories, and this removal simplifies the > code logic in several places. > > +If one of your recipes is still using this mechanism, you can split the > +code source fetching into two separate entries:: > + > + SRC_URI = "git://some.host/somepath;branch=branchX;name=nameX \ > + git://some.host/somepath;branch=branchY;name=nameY" > + SRCREV_nameX = "xxxxxxxxxxxxxxxxxxxx" > + SRCREV_nameY = "yyyyyyyyyyyyyyyyyyyy" > + > Git fetcher: Branch parameter now required in :term:`SRC_URI` > ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ > > @@ -242,7 +250,7 @@ The following recipes have been removed in this release: > :oe_git:`openembedded-core </openembedded-core>` or :oe_git:`meta-openembedded </meta-openembedded>`. > > - ``blktool``: It was created in 2004 as an alternative to hdparm and never > - updated since (while :wikipedia:`hdparm <Hdparm>` remains in active. > + updated since (while :wikipedia:`hdparm <Hdparm>` remains in active). Mmmm, should we say s/remains in active/remains active/ ? Slow brain day so maybe my intuition is wrong today :) > > - ``cargo-c-native``: converted to a target recipe and renamed to ``cargo-c``. > > diff --git a/documentation/migration-guides/release-notes-5.2.rst b/documentation/migration-guides/release-notes-5.2.rst > index 05fc92e85..2ad441449 100644 > --- a/documentation/migration-guides/release-notes-5.2.rst > +++ b/documentation/migration-guides/release-notes-5.2.rst > @@ -44,11 +44,10 @@ New Features / Enhancements in |yocto-ver| > uses. > > - The :term:`GRUB_MKIMAGE_OPTS` can be used to control the flags to the > - ``grub-mkimage`` command in the context of the Grub recipe (``grub-efi``). > + ``grub-mkimage`` command in the context of the GRUB recipe (``grub-efi``). > > - The :term:`SPDX_PACKAGE_VERSION` variable controls the package version as > - seen in the SPDX 3.0 JSON output (``software_packageVersion``). The default > - value for this variable is :term:`PV`. > + seen in the SPDX 3.0 JSON output (``software_packageVersion``). The indentation seems odd here (but to be fair, it is also in the part of the diff that is being removed). Cheers, Quentin
Hi Quentin, On Thu Apr 3, 2025 at 4:23 PM CEST, Quentin Schulz wrote: > Hi Antonin, > > On 4/3/25 4:08 PM, Antonin Godard wrote: >> Address some of the reviews received on the release notes and migration >> guide for 5.2: >> >> - Remove the wget fetcher addition, I misread commit 0a9f90ff658e >> ("tests/fetch: support setting PV in the wget fetcher") from BitBake, >> and it actually affects the tests for the wget fetcher, not the >> fetcher itself. >> - Add a way to transition from the branch=nameX,nameY syntax. From what >> I have tested, I achieve the same behavior when using one or the other >> syntax. >> - Remove the default value for SPDX_PACKAGE_VERSION in the release note. >> - Fix a couple of typos. >> >> Suggested-by: Quentin Schulz <quentin.schulz@cherry.de> >> Signed-off-by: Antonin Godard <antonin.godard@bootlin.com> >> --- >> documentation/migration-guides/migration-5.2.rst | 10 +++++++++- >> documentation/migration-guides/release-notes-5.2.rst | 7 ++----- >> 2 files changed, 11 insertions(+), 6 deletions(-) >> >> diff --git a/documentation/migration-guides/migration-5.2.rst b/documentation/migration-guides/migration-5.2.rst >> index c54f054c7..59f21ff4c 100644 >> --- a/documentation/migration-guides/migration-5.2.rst >> +++ b/documentation/migration-guides/migration-5.2.rst >> @@ -193,6 +193,14 @@ anymore:: >> This was rarely used in the core repositories, and this removal simplifies the >> code logic in several places. >> >> +If one of your recipes is still using this mechanism, you can split the >> +code source fetching into two separate entries:: >> + >> + SRC_URI = "git://some.host/somepath;branch=branchX;name=nameX \ >> + git://some.host/somepath;branch=branchY;name=nameY" >> + SRCREV_nameX = "xxxxxxxxxxxxxxxxxxxx" >> + SRCREV_nameY = "yyyyyyyyyyyyyyyyyyyy" >> + >> Git fetcher: Branch parameter now required in :term:`SRC_URI` >> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ >> >> @@ -242,7 +250,7 @@ The following recipes have been removed in this release: >> :oe_git:`openembedded-core </openembedded-core>` or :oe_git:`meta-openembedded </meta-openembedded>`. >> >> - ``blktool``: It was created in 2004 as an alternative to hdparm and never >> - updated since (while :wikipedia:`hdparm <Hdparm>` remains in active. >> + updated since (while :wikipedia:`hdparm <Hdparm>` remains in active). > > Mmmm, should we say > > s/remains in active/remains active/ ? > > Slow brain day so maybe my intuition is wrong today :) No, the slowness seems to have been on my side the other day, thanks ;) Will fix. >> >> - ``cargo-c-native``: converted to a target recipe and renamed to ``cargo-c``. >> >> diff --git a/documentation/migration-guides/release-notes-5.2.rst b/documentation/migration-guides/release-notes-5.2.rst >> index 05fc92e85..2ad441449 100644 >> --- a/documentation/migration-guides/release-notes-5.2.rst >> +++ b/documentation/migration-guides/release-notes-5.2.rst >> @@ -44,11 +44,10 @@ New Features / Enhancements in |yocto-ver| >> uses. >> >> - The :term:`GRUB_MKIMAGE_OPTS` can be used to control the flags to the >> - ``grub-mkimage`` command in the context of the Grub recipe (``grub-efi``). >> + ``grub-mkimage`` command in the context of the GRUB recipe (``grub-efi``). >> >> - The :term:`SPDX_PACKAGE_VERSION` variable controls the package version as >> - seen in the SPDX 3.0 JSON output (``software_packageVersion``). The default >> - value for this variable is :term:`PV`. >> + seen in the SPDX 3.0 JSON output (``software_packageVersion``). > > The indentation seems odd here (but to be fair, it is also in the part > of the diff that is being removed). Hmm, already on the previous patches you noticed this I think, but after verifying in the code, the alignment was proper. Actually, I may have already seen this in your answers before, I wonder if the problem is on my side or yours. Exporting the email in an mbox file does not show any issue as well. Antonin
Hi Antonin, On 4/3/25 5:10 PM, Antonin Godard wrote: > Hi Quentin, > > On Thu Apr 3, 2025 at 4:23 PM CEST, Quentin Schulz wrote: >> Hi Antonin, >> >> On 4/3/25 4:08 PM, Antonin Godard wrote: >>> Address some of the reviews received on the release notes and migration >>> guide for 5.2: >>> >>> - Remove the wget fetcher addition, I misread commit 0a9f90ff658e >>> ("tests/fetch: support setting PV in the wget fetcher") from BitBake, >>> and it actually affects the tests for the wget fetcher, not the >>> fetcher itself. >>> - Add a way to transition from the branch=nameX,nameY syntax. From what >>> I have tested, I achieve the same behavior when using one or the other >>> syntax. >>> - Remove the default value for SPDX_PACKAGE_VERSION in the release note. >>> - Fix a couple of typos. >>> >>> Suggested-by: Quentin Schulz <quentin.schulz@cherry.de> >>> Signed-off-by: Antonin Godard <antonin.godard@bootlin.com> >>> --- >>> documentation/migration-guides/migration-5.2.rst | 10 +++++++++- >>> documentation/migration-guides/release-notes-5.2.rst | 7 ++----- >>> 2 files changed, 11 insertions(+), 6 deletions(-) >>> >>> diff --git a/documentation/migration-guides/migration-5.2.rst b/documentation/migration-guides/migration-5.2.rst >>> index c54f054c7..59f21ff4c 100644 >>> --- a/documentation/migration-guides/migration-5.2.rst >>> +++ b/documentation/migration-guides/migration-5.2.rst >>> @@ -193,6 +193,14 @@ anymore:: >>> This was rarely used in the core repositories, and this removal simplifies the >>> code logic in several places. >>> >>> +If one of your recipes is still using this mechanism, you can split the >>> +code source fetching into two separate entries:: >>> + >>> + SRC_URI = "git://some.host/somepath;branch=branchX;name=nameX \ >>> + git://some.host/somepath;branch=branchY;name=nameY" >>> + SRCREV_nameX = "xxxxxxxxxxxxxxxxxxxx" >>> + SRCREV_nameY = "yyyyyyyyyyyyyyyyyyyy" >>> + Just to make sure we don't miss this for the v2, but this "solution" was rejected by Richard as non-working, see https://lore.kernel.org/yocto-docs/9c3af54f0c74af0b9d0887e0408fd591e936c4da.camel@linuxfoundation.org/T/#u Cheers, Quentin >>> Git fetcher: Branch parameter now required in :term:`SRC_URI` >>> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ >>> >>> @@ -242,7 +250,7 @@ The following recipes have been removed in this release: >>> :oe_git:`openembedded-core </openembedded-core>` or :oe_git:`meta-openembedded </meta-openembedded>`. >>> >>> - ``blktool``: It was created in 2004 as an alternative to hdparm and never >>> - updated since (while :wikipedia:`hdparm <Hdparm>` remains in active. >>> + updated since (while :wikipedia:`hdparm <Hdparm>` remains in active). >> >> Mmmm, should we say >> >> s/remains in active/remains active/ ? >> >> Slow brain day so maybe my intuition is wrong today :) > > No, the slowness seems to have been on my side the other day, thanks ;) Will fix. > >>> >>> - ``cargo-c-native``: converted to a target recipe and renamed to ``cargo-c``. >>> >>> diff --git a/documentation/migration-guides/release-notes-5.2.rst b/documentation/migration-guides/release-notes-5.2.rst >>> index 05fc92e85..2ad441449 100644 >>> --- a/documentation/migration-guides/release-notes-5.2.rst >>> +++ b/documentation/migration-guides/release-notes-5.2.rst >>> @@ -44,11 +44,10 @@ New Features / Enhancements in |yocto-ver| >>> uses. >>> >>> - The :term:`GRUB_MKIMAGE_OPTS` can be used to control the flags to the >>> - ``grub-mkimage`` command in the context of the Grub recipe (``grub-efi``). >>> + ``grub-mkimage`` command in the context of the GRUB recipe (``grub-efi``). >>> >>> - The :term:`SPDX_PACKAGE_VERSION` variable controls the package version as >>> - seen in the SPDX 3.0 JSON output (``software_packageVersion``). The default >>> - value for this variable is :term:`PV`. >>> + seen in the SPDX 3.0 JSON output (``software_packageVersion``). >> >> The indentation seems odd here (but to be fair, it is also in the part >> of the diff that is being removed). > > Hmm, already on the previous patches you noticed this I think, but after > verifying in the code, the alignment was proper. Actually, I may have already > seen this in your answers before, I wonder if the problem is on my side or > yours. Exporting the email in an mbox file does not show any issue as well. > Checked myself in the tree directly, indeed, 3 whitespaces followed by a dash, then two whitespaces and then "The". The next line is starting after 6 whitespaces so the indentation is proper. I looked on lore.kernel.org and it seems like your patch was fine, and then looking into my answer, it's oddly indented again. Maybe my mail client doing something, or my mail server doing weird things again. Please ignore then :) Cheers, Quentin
Hi Quentin, On Thu Apr 3, 2025 at 7:37 PM CEST, Quentin Schulz wrote: > Hi Antonin, > > On 4/3/25 5:10 PM, Antonin Godard wrote: >> Hi Quentin, >> >> On Thu Apr 3, 2025 at 4:23 PM CEST, Quentin Schulz wrote: >>> Hi Antonin, >>> >>> On 4/3/25 4:08 PM, Antonin Godard wrote: >>>> Address some of the reviews received on the release notes and migration >>>> guide for 5.2: >>>> >>>> - Remove the wget fetcher addition, I misread commit 0a9f90ff658e >>>> ("tests/fetch: support setting PV in the wget fetcher") from BitBake, >>>> and it actually affects the tests for the wget fetcher, not the >>>> fetcher itself. >>>> - Add a way to transition from the branch=nameX,nameY syntax. From what >>>> I have tested, I achieve the same behavior when using one or the other >>>> syntax. >>>> - Remove the default value for SPDX_PACKAGE_VERSION in the release note. >>>> - Fix a couple of typos. >>>> >>>> Suggested-by: Quentin Schulz <quentin.schulz@cherry.de> >>>> Signed-off-by: Antonin Godard <antonin.godard@bootlin.com> >>>> --- >>>> documentation/migration-guides/migration-5.2.rst | 10 +++++++++- >>>> documentation/migration-guides/release-notes-5.2.rst | 7 ++----- >>>> 2 files changed, 11 insertions(+), 6 deletions(-) >>>> >>>> diff --git a/documentation/migration-guides/migration-5.2.rst b/documentation/migration-guides/migration-5.2.rst >>>> index c54f054c7..59f21ff4c 100644 >>>> --- a/documentation/migration-guides/migration-5.2.rst >>>> +++ b/documentation/migration-guides/migration-5.2.rst >>>> @@ -193,6 +193,14 @@ anymore:: >>>> This was rarely used in the core repositories, and this removal simplifies the >>>> code logic in several places. >>>> >>>> +If one of your recipes is still using this mechanism, you can split the >>>> +code source fetching into two separate entries:: >>>> + >>>> + SRC_URI = "git://some.host/somepath;branch=branchX;name=nameX \ >>>> + git://some.host/somepath;branch=branchY;name=nameY" >>>> + SRCREV_nameX = "xxxxxxxxxxxxxxxxxxxx" >>>> + SRCREV_nameY = "yyyyyyyyyyyyyyyyyyyy" >>>> + > > Just to make sure we don't miss this for the v2, but this "solution" was > rejected by Richard as non-working, see > https://lore.kernel.org/yocto-docs/9c3af54f0c74af0b9d0887e0408fd591e936c4da.camel@linuxfoundation.org/T/#u Hm, but there are recipes using this in oe-core like [1]. Does that mean that these are broken at the moment? In my testing, I could properly fetch and unpack such recipes with one syntax or the other. Indeed, after unpacking, only the source code of the first item of SRC_URI is unpacked and present in the WORKDIR, yet BitBake does not complain about the second one or produce any kind of error. So I'm not sure of what to conclude on this? I'm not sure what is the best to say in the release notes. [1]: https://git.openembedded.org/openembedded-core/tree/meta-selftest/recipes-extended/sysdig/sysdig-selftest_0.28.0.bb#n25 Antonin
Hi Antonin, On 4/4/25 11:34 AM, Antonin Godard wrote: > Hi Quentin, > > On Thu Apr 3, 2025 at 7:37 PM CEST, Quentin Schulz wrote: >> Hi Antonin, >> >> On 4/3/25 5:10 PM, Antonin Godard wrote: >>> Hi Quentin, >>> >>> On Thu Apr 3, 2025 at 4:23 PM CEST, Quentin Schulz wrote: >>>> Hi Antonin, >>>> >>>> On 4/3/25 4:08 PM, Antonin Godard wrote: >>>>> Address some of the reviews received on the release notes and migration >>>>> guide for 5.2: >>>>> >>>>> - Remove the wget fetcher addition, I misread commit 0a9f90ff658e >>>>> ("tests/fetch: support setting PV in the wget fetcher") from BitBake, >>>>> and it actually affects the tests for the wget fetcher, not the >>>>> fetcher itself. >>>>> - Add a way to transition from the branch=nameX,nameY syntax. From what >>>>> I have tested, I achieve the same behavior when using one or the other >>>>> syntax. >>>>> - Remove the default value for SPDX_PACKAGE_VERSION in the release note. >>>>> - Fix a couple of typos. >>>>> >>>>> Suggested-by: Quentin Schulz <quentin.schulz@cherry.de> >>>>> Signed-off-by: Antonin Godard <antonin.godard@bootlin.com> >>>>> --- >>>>> documentation/migration-guides/migration-5.2.rst | 10 +++++++++- >>>>> documentation/migration-guides/release-notes-5.2.rst | 7 ++----- >>>>> 2 files changed, 11 insertions(+), 6 deletions(-) >>>>> >>>>> diff --git a/documentation/migration-guides/migration-5.2.rst b/documentation/migration-guides/migration-5.2.rst >>>>> index c54f054c7..59f21ff4c 100644 >>>>> --- a/documentation/migration-guides/migration-5.2.rst >>>>> +++ b/documentation/migration-guides/migration-5.2.rst >>>>> @@ -193,6 +193,14 @@ anymore:: >>>>> This was rarely used in the core repositories, and this removal simplifies the >>>>> code logic in several places. >>>>> >>>>> +If one of your recipes is still using this mechanism, you can split the >>>>> +code source fetching into two separate entries:: >>>>> + >>>>> + SRC_URI = "git://some.host/somepath;branch=branchX;name=nameX \ >>>>> + git://some.host/somepath;branch=branchY;name=nameY" >>>>> + SRCREV_nameX = "xxxxxxxxxxxxxxxxxxxx" >>>>> + SRCREV_nameY = "yyyyyyyyyyyyyyyyyyyy" >>>>> + >> >> Just to make sure we don't miss this for the v2, but this "solution" was >> rejected by Richard as non-working, see >> https://lore.kernel.org/yocto-docs/9c3af54f0c74af0b9d0887e0408fd591e936c4da.camel@linuxfoundation.org/T/#u > > Hm, but there are recipes using this in oe-core like [1]. Does that mean that > these are broken at the moment? > > In my testing, I could properly fetch and unpack such recipes with one syntax or > the other. Indeed, after unpacking, only the source code of the first item of > SRC_URI is unpacked and present in the WORKDIR, yet BitBake does not complain > about the second one or produce any kind of error. So I'm not sure of what to > conclude on this? > I think what Richard meant was that it's brittle/finicky/unstable to use twice the same git repo in SRC_URI but with a different branch/srcrev? And that it's believed nobody did need that behavior (only the kernel recipe was using that in OE-Core, ever). > I'm not sure what is the best to say in the release notes. > I guess he suggests to simply not document how to fix it because it's "incorrect" (my understanding of what he said) and virtually nobody was impacted. Cheers, Quentin
diff --git a/documentation/migration-guides/migration-5.2.rst b/documentation/migration-guides/migration-5.2.rst index c54f054c7..59f21ff4c 100644 --- a/documentation/migration-guides/migration-5.2.rst +++ b/documentation/migration-guides/migration-5.2.rst @@ -193,6 +193,14 @@ anymore:: This was rarely used in the core repositories, and this removal simplifies the code logic in several places. +If one of your recipes is still using this mechanism, you can split the +code source fetching into two separate entries:: + + SRC_URI = "git://some.host/somepath;branch=branchX;name=nameX \ + git://some.host/somepath;branch=branchY;name=nameY" + SRCREV_nameX = "xxxxxxxxxxxxxxxxxxxx" + SRCREV_nameY = "yyyyyyyyyyyyyyyyyyyy" + Git fetcher: Branch parameter now required in :term:`SRC_URI` ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ @@ -242,7 +250,7 @@ The following recipes have been removed in this release: :oe_git:`openembedded-core </openembedded-core>` or :oe_git:`meta-openembedded </meta-openembedded>`. - ``blktool``: It was created in 2004 as an alternative to hdparm and never - updated since (while :wikipedia:`hdparm <Hdparm>` remains in active. + updated since (while :wikipedia:`hdparm <Hdparm>` remains in active). - ``cargo-c-native``: converted to a target recipe and renamed to ``cargo-c``. diff --git a/documentation/migration-guides/release-notes-5.2.rst b/documentation/migration-guides/release-notes-5.2.rst index 05fc92e85..2ad441449 100644 --- a/documentation/migration-guides/release-notes-5.2.rst +++ b/documentation/migration-guides/release-notes-5.2.rst @@ -44,11 +44,10 @@ New Features / Enhancements in |yocto-ver| uses. - The :term:`GRUB_MKIMAGE_OPTS` can be used to control the flags to the - ``grub-mkimage`` command in the context of the Grub recipe (``grub-efi``). + ``grub-mkimage`` command in the context of the GRUB recipe (``grub-efi``). - The :term:`SPDX_PACKAGE_VERSION` variable controls the package version as - seen in the SPDX 3.0 JSON output (``software_packageVersion``). The default - value for this variable is :term:`PV`. + seen in the SPDX 3.0 JSON output (``software_packageVersion``). - Kernel-related changes: @@ -345,8 +344,6 @@ New Features / Enhancements in |yocto-ver| - ``wget``: increase timeout to 100s from 30s to match CDN worst response time. - - ``wget``: Support setting :term:`PV` in :term:`SRC_URI`. - - Add support for fast initial shallow fetch. The fetcher will prefer an initial shallow clone, but will re-utilize an existing bare clone if there is one. If the remote server does not allow shallow fetches, the
Address some of the reviews received on the release notes and migration guide for 5.2: - Remove the wget fetcher addition, I misread commit 0a9f90ff658e ("tests/fetch: support setting PV in the wget fetcher") from BitBake, and it actually affects the tests for the wget fetcher, not the fetcher itself. - Add a way to transition from the branch=nameX,nameY syntax. From what I have tested, I achieve the same behavior when using one or the other syntax. - Remove the default value for SPDX_PACKAGE_VERSION in the release note. - Fix a couple of typos. Suggested-by: Quentin Schulz <quentin.schulz@cherry.de> Signed-off-by: Antonin Godard <antonin.godard@bootlin.com> --- documentation/migration-guides/migration-5.2.rst | 10 +++++++++- documentation/migration-guides/release-notes-5.2.rst | 7 ++----- 2 files changed, 11 insertions(+), 6 deletions(-)