Age | Commit message (Collapse) | Author |
|
Generated from commits in the kirkstone branch, as well as a few entries
from the migration guide.
(Note that the "Repositories / Downloads" section still needs to be
filled in.)
Signed-off-by: Paul Eggleton <paul.eggleton@microsoft.com>
|
|
* Make some corrections to preliminary set of entries
* Move out entries that were more appropriate for the release notes
(i.e. that are more additions rather than changes that require the
user to make changes)
* Add new entries based on commits in the kirkstone branch
Signed-off-by: Paul Eggleton <paul.eggleton@microsoft.com>
|
|
This check is new in kirkstone.
Signed-off-by: Paul Eggleton <paul.eggleton@microsoft.com>
|
|
ZSTD_THREADS is new for kirkstone, XZ_THREADS was introduced in dunfell.
Signed-off-by: Paul Eggleton <paul.eggleton@microsoft.com>
|
|
New variable in kirkstone.
Signed-off-by: Paul Eggleton <paul.eggleton@microsoft.com>
|
|
Mention the vendor filtering functionality - prompted by OE-Core
revision 45d1a0bea0c628f84a00d641a4d323491988106f.
Signed-off-by: Paul Eggleton <paul.eggleton@microsoft.com>
|
|
Add this since the INCOMPATIBLE_LICENSE wildcard support has changed in
the 4.0 release.
Signed-off-by: Paul Eggleton <paul.eggleton@microsoft.com>
|
|
It seems prudent to point out that hard-coding passwords in the manner
detailed in the example is not a good idea in production. This type of
mistake has unfortunately been made by many device vendors (outside of
the Yocto Project context) leading to security vulnerabilities.
Signed-off-by: Paul Eggleton <paul.eggleton@microsoft.com>
|
|
We missed noting this in 3.4 but I noticed the documentation was
recently updated, so note the removal.
Signed-off-by: Paul Eggleton <paul.eggleton@microsoft.com>
|
|
Branches can never be outdated so let's check the release is not a
branch (branches end with .999).
Cc: Quentin Schulz <foss+yocto@0leil.net>
Signed-off-by: Quentin Schulz <quentin.schulz@theobroma-systems.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The switcher expects URL subpath to match the "release" used by sphinx
to build the documentation. Branches, however, are put in a subpath
after their name (e.g. dunfell) while sphinx sets the "release" to
X.Y.999. This means the switcher cannot replace correctly the path to
switch between releases/versions.
Let set_versions.py inject the list of release names into the
switchers.js.in file so it can check whether the subpath is one of the
release names in which case it needs to be stripped.
Cc: Quentin Schulz <foss+yocto@0leil.net>
Signed-off-by: Quentin Schulz <quentin.schulz@theobroma-systems.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
This array only contains versions that can be selected through the
dropdown menu for switching between supported versions.
Therefore, let's rename it to switcher_versions to make its usecase
clearer.
Cc: Quentin Schulz <foss+yocto@0leil.net>
Signed-off-by: Quentin Schulz <quentin.schulz@theobroma-systems.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
obsolete releases
Branches are identified by their .999 version suffix which means they
will never be matched in the forloop above this git context. Therefore,
branches will match the condition. However, branches are not necessarily
obsolete (e.g. dunfell, honister and kirkstone today), so let's mark as
obsolete the branches which are from obsolete releases.
Old tags of currently supported releases are not defined as obsolete but
outdated, therefore using the series to which they belong like it is
done for branches is enough for obsolescence detection.
Cc: Quentin Schulz <foss+yocto@0leil.net>
Signed-off-by: Quentin Schulz <quentin.schulz@theobroma-systems.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
master branch of Bitbake is now located at docs.yoctoproject.org/dev
instead of docs.yoctoproject.org so let's update the switchers and
set_versions.py to reflect that change.
Cc: Quentin Schulz <foss+yocto@0leil.net>
Signed-off-by: Quentin Schulz <quentin.schulz@theobroma-systems.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
ourseries can be an active release and therefore shouldn't be marked as
obsolete. By adding ourseries to activereleases, it is impossible to
know if ourseries is actually an active release or not. Instead let's
loop on the active releases with ourseries too (only if it's not active
release, otherwise it'd appear twice).
Fixes: 6f40ef56054ec "docs: set_versions.py: add information about obsolescence of a release"
Cc: Quentin Schulz <foss+yocto@0leil.net>
Signed-off-by: Quentin Schulz <quentin.schulz@theobroma-systems.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
switchers.js
versions array is supposed to store the latest version of all active
releases. However, in the loop it is reassigned and therefore, the check
on whether our version is already in the versions array will always
return false (except for the latest version of the last active release)
and write our version again in the list.
By using a local variable for the logic instead of versions array, the
check now works properly.
Fixes: f2b069be8c307 "set_versions: Various improvements"
Cc: Quentin Schulz <foss+yocto@0leil.net>
Signed-off-by: Quentin Schulz <quentin.schulz@theobroma-systems.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Based on additional information per release, specifically with the
obsolescence status of a release, the obsolescence detection can now be
much smarter than just checking if the release is older than dunfell.
This is required because with LTS (dunfell for example) releases, it is
now possible to have LTS releases that are older than obsolete releases.
This means obsolete releases need to be tracked and only the release
version cannot be used as an indicator of obsolescence.
Let's use the obsolete field of the per-release data in the all_versions
dictionary to display correct warning messages.
The warning message is first about outdated version if there's a newer
one available (*even* if it is for an obsolete release, e.g. 3.0.1 will
say it's outdated and should select 3.0.4 version instead), then if the
version is the last of the release, show a warning message if the
release is obsolete.
Cc: Quentin Schulz <foss+yocto@0leil.net>
Signed-off-by: Quentin Schulz <quentin.schulz@theobroma-systems.com>
|
|
This adds support for marking releases as obsolete to make the
detection algorithm smarter (in a later commit) than just checking if
it's older than dunfell.
Cc: Quentin Schulz <foss+yocto@0leil.net>
Signed-off-by: Quentin Schulz <quentin.schulz@theobroma-systems.com>
|
|
Since commit f2b069be8c307 "set_versions: Various improvements", an
outdated version will always appear in all_versions, meaning there'll
always be an exact match in the loop (just above the git context of this
patch) so there's no need to add the current_version to the dropdown
menu manually.
This issue showed up only for outdated versions of obsolete releases,
e.g. 3.2.3. In that case, 3.2.4 (latest version of the obsolete release)
will appear in the all_versions array in addition to 3.2.3, which means
the check on release series (3.2) will be matched twice, and 3.2.3 will
be printed once in the 3.2.4 loop because version != current_version and
once in the 3.2.3 because it is an exact match to an entry in
all_versions.
Cc: Quentin Schulz <foss+yocto@0leil.net>
Signed-off-by: Quentin Schulz <quentin.schulz@theobroma-systems.com>
|
|
Signed-off-by: Abongwa Bonalais Amahnui <abongwabonalais@gmail.com>
Reviewed-by: Michael Opdenacker <michael.opdenacker@bootlin.com>
|
|
Replacing local.conf by conf/local.conf
Signed-off-by: Abongwa Bonalais Amahnui <abongwabonalais@gmail.com>
Reviewed-by: Michael Opdenacker <michael.opdenacker@bootlin.com>
|
|
Update the docs to reflect that kirkstone is 4.0, not 3.5.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Signed-off-by: Michael Opdenacker <michael.opdenacker@bootlin.com>
|
|
Otherwise rendered as "en" dashes by sphinx, for example:
git pull –rebase
instead of
git pull --rebase
Readers could confuse them with normal dashes.
Exception: replace by \-\- in italic text (*text*)
as Sphinx cannot nest italic and quoted text.
For consistency, also update quoting in strings in the same code hunks.
Signed-off-by: Michael Opdenacker <michael.opdenacker@bootlin.com>
|
|
Signed-off-by: Michael Opdenacker <michael.opdenacker@bootlin.com>
|
|
Signed-off-by: Michael Opdenacker <michael.opdenacker@bootlin.com>
|
|
Add the list of renamed and removed variables, show how the renaming
script works. Also mention the change in cve-check output.
Signed-off-by: Marta Rybczynska <marta.rybczynska@huawei.com>
Reviewed-by: Michael Opdenacker <michael.opdenacker@bootlin.com>
|
|
Signed-off-by: Daniel Ammann <daniel.ammann@bytesatwork.ch>
Reviewed-by: Michael Opdenacker <michael.opdenacker@bootlin.com>
|
|
Signed-off-by: Daniel Ammann <daniel.ammann@bytesatwork.ch>
Reviewed-by: Michael Opdenacker <michael.opdenacker@bootlin.com>
|
|
The AVAILABLE_LICENSES variable has been removed from OE-Core.
Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Reviewed-by: Michael Opdenacker <michael.opdenacker@bootlin.com>
|
|
Add new PEP517_* variables, remove obsolete variables, and update for
renamed classes.
Signed-off-by: Ross Burton <ross.burton@arm.com>
Reviewed-by: Michael Opdenacker <michael.opdenacker@bootlin.com>
|
|
I believe we should only be showing the "active" versions in the switchers
base list of releases to show. Zeus and warrior are old and no longer actively
maintained and we don't suggest new users use them. gatesgarth is also
outside it's support window.
I therefore propose removing these leaving us with dev, honister, hardknott,
dunfell (LTS). In addition, any release that is selected will be shown, along
with any later release in that release series.
People can still navigate to the older docs using the releases list or by
direct URL but this highlights to users which releases we'd expect/encourage
them to be using.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Build scripts now only require updates in master.
Signed-off-by: Michael Halstead <mhalstead@linuxfoundation.org>
|
|
Correct the previous commit since we need to compare against ourseries
rather than ourbranch since devbranch would match against the series,
not the branch.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The dev branch is being displayed in switchers under two different
headings, 4.0.999 and dev (4.0). Add an additional conditional to fix that.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
- Allow specifying the version from the commandline
- Add all previous release series/version mappings (to support transitions branch)
- Add poky mapping for 3.4 as some releases erronously use it
- Improve git branch 'guessing' code to work properly
- Handle poky '.0' release mappings correctly
- Only write poky.yaml if poky.yaml.in exists
- Ensure older non-active releases are shown in the switchers.js
release list.
- Ensure current version and current release series are shown in switcher.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
autogenerated
A horrible blunt hammer approach to updating the version information in
switchers.js based on the available tag information.
To merge and work correctly, this will need a change to the autobuilder-helper
docs generation code to pull the swicthers.js and script from master, then
to run the script. That should hopefully remove the need for other patching
even on old docs branches though.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Allow conf.py to read the versions it needs from poky.yaml and have
set_versions.py write this out. This means we don't have to change as
many files when making new releases.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Use a script to generate the branch/tag information inside poky.yaml.
If the branch isn't a known release branch, include git magic to find
the closest matching release branch we know about.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Adding 3.1.15 to documentation switcher and release list.
Signed-off-by: Michael Halstead <mhalstead@linuxfoundation.org>
|
|
This doesn't seem to be used anymore and doesn't make much sense,
remove it.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Reviewed-by: Michael Opdenacker <michael.opdenacker@bootlin.com>
|
|
We don't publish tarballs using this version any more, it has been
removed everywhere. Remove the obsolete references in the docs,
adapting references accordingly. The tarballs now are just created
with their sha256 checksum which we don't want or need to reference
in the docs.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Reviewed-by: Michael Opdenacker <michael.opdenacker@bootlin.com>
|
|
Signed-off-by: Michael Opdenacker <michael.opdenacker@bootlin.com>
|
|
Signed-off-by: Michael Opdenacker <michael.opdenacker@bootlin.com>
|
|
journald is not a syslog provider, so using journald exclusively amounts
to just preventing a syslog provider from being specified on a system
where journald is active.
Various spots where this has popped up:
* https://stackoverflow.com/q/48746397
* https://lists.yoctoproject.org/g/yocto/topic/82883035#53548
Signed-off-by: Nicholas Sielicki <nicholas.sielicki@legrand.com>
|
|
By alphabetical order, to get the same order as in the HTML
output, sorted thanks to the ":sorted:" directive.
That was needed after the recent variable renaming for
inclusive language, but other variables were already
out of order anyway.
Signed-off-by: Michael Opdenacker <michael.opdenacker@bootlin.com>
---
The change was made manually, but the source order was compared automatically
with the generated one thanks to these quick and dirty commands...
To extract the unsorted list of variable definitions from the sources:
> git grep " :term:" documentation/ref-manual/variables.rst | grep -v " " | cut -d ":" -f4 | cut -d "\`" -f2 > ~/tmp/unsorted
To extract the sorted list of variables from the generated HTML output:
> cd documentation
> make html
> egrep -o "^<dt id=\"term-([0-9A-Za-z\-_])*" _build/html/ref-manual/variables.html | cut -c 14- > ~/tmp/sorted
You'll most probably have better and simpler ideas for generating such lists!
|
|
As already done in the BitBake manual.
Even though we're trying to keep the variable
definitions in alphabetical order, it's useful
to make sure that the variables are ordered in the
generated output.
Signed-off-by: Michael Opdenacker <michael.opdenacker@bootlin.com>
Reported-by: Quentin Schulz <foss@0leil.net>
|
|
Add AlmaLinux 8.5, Fedora 35, and OpenSUSE Leap 15.3
Remove Fedora 33, Debian 8, OpenSUSE Leap 15.1 and 15.2
Removing distros that were recently removed from SANITY_TESTED_DISTROS
in meta-poky/conf/distro/poky.conf
Signed-off-by: Michael Opdenacker <michael.opdenacker@bootlin.com>
Reported-by: Steve Sakoman <steve@sakoman.com>
Reported-by: Christian Eggers <ceggers@arri.de>
|
|
A number of classes and variables have been renamed, so update the
documentation as needed.
Signed-off-by: Ross Burton <ross.burton@arm.com>
Reviewed-by: Michael Opdenacker <michael.opdenacker@bootlin.com>
|