Extra-data usually is downloaded on the user side into an ostree repo.
For system installs, a temporary ostree repo is used on the user side
and then imported on the system side. This doesn't work for OCI images
because importing the image into an ostree repo makes it impossible for
the system side to verify the data.
So instead, the OCI image is first mirrored into a local OCI repo and
then gets imported on the system side, which can verify the image from
the index by the digest.
Closes: https://github.com/flatpak/flatpak/issues/3790
It took in a ostree commit before, but it really only needs the detached
metadata. The single caller now converts it correctly.
This will become useful when we want to know the size of the extra data
when the metadata is coming from an OCI image.
This cleans up flatpak_dir_pull_extra_data in a way that it is only
responsible for downloading the extra-data. In has no dependencies on
ostree concepts any more.
The new flatpak_dir_pull_ostree_extra_data now takes care of extracing
the extra data sources and saving it in the detached metadata.
CURLcode return values from curl_easy_setopt() were being compared
with CURLM_OK (a CURLMcode constant) instead of CURLE_OK. This is
incorrect as CURLM_OK is for multi interface functions, not easy
interface functions.
A few years ago there was a very painful attempt at porting from
libsoup2 to libsoup3. Flatpak libsoup3 support never landed and it seems
like a large amount of distros have switched over to libcurl instead.
This commit removes libsoup2 support completely instead of growing
libsoup3 support.
Closes#5915Closes#4582
Add an option to build OCI bundles with zstd compressed layers.
gzip is kept as the default for maximum compatibility:
Ecosystem support:
distribution/distribution: no explicit support, but works
quay.io: sinc 2021
Amazon ECR: supported
pulp_container: since 2022
flatpak: since first-OCI supporting version
tardiff: since first version
- Make percent values translatable
Various languages use different ways to format the percentage
values[1], making it translatable will allow a more coherent way to
display the information.
- Make remaining time abbreviation translatable
Making this value translatable will allow languages to display the
seconds abbreviation in their language. This is particulary an issue
for Turkish, in which hours and seconds start with the same letter, so
it's not possible to distinguish which is which. We use a second
letter (sa, sn) to figure it out.
[1] https://en.wikipedia.org/wiki/Percent_sign#Form_and_spacing
Signed-off-by: Emir SARI <emir_sari@icloud.com>
In 4febfb59 ("flatpak: Disable progress escape sequence by default") the
escape sequence has been disabled by default, but we want to enable it
again for 1.18.
This comment needs to be modernized for 2025. It notably fails to
mention the two projects that are actually still using code that looks
like Flatpak's.
We require jobs to succeed before a PR can be merged. If a PR only
touches files which match `paths-ignore`, the jobs won't run and the PR
can't be merged.
The number of PRs which only touch those files is small, and the extra
CI runs resulting from that can be ignored.
If an SDK is already installed in a dir that is not targeted with a
flatpak transaction, and the transaction has auto_install_sdk set,
add_new_dep_op returns NULL in dep_op which is not correctly handled in
add_deps.
Fixes#5894
There are a number of races, and failure conditions which can lead to a
pid of 0 being returned from flatpak_instance_get_child_pid. This would
lead to a whole bunch of things getting killed.
We will skip the instance in those cases now, and retry a few times. We
also notice when the instance just goes away by itself now.
This should make killing more robust, and especially not SIGKILL pid 0.
We could previously just assume that the first enabled remote
(potentially matching the collection ID) contains the ref, but that
obviously is not always the case.
The change here looks up the remote state of the remotes to figure out
if they actually contain the ref, and adds the first matching remote to
the transaction.
Sudo can be used in several ways other than calling a command with the
root user. For example, one can use -u to run the command as the
specified user, or -g to specify a primary group to run the command
as.
Flatpak adds a check when --user is used to prevent an installation in
the root's directory, for example, but it does it by only checking if
sudo was used. As stated previously, it does not necessarily imply
root, so this patch explicitly checks if the command is being run with
the root user.
Fixes: https://github.com/flatpak/flatpak/issues/5979
Signed-off-by: Georgia Garcia <georgia.garcia@canonical.com>
Those schemes are in use by flathub, usually to be handled by an app
store, but we can also handle it directly with `flatpak install`.
Base on an idea from user bbb651 in
https://github.com/flatpak/flatpak/pull/6259.
The appstream docs say:
Release data may be present directly in a component metainfo file, but
also optionally be split out into an external metadata file.
The file must be installed as
/usr/share/metainfo/releases/%{cid}.releases.xml, where cid is the
component ID of the component the release information belongs to.
This change allows flatpak build-finish to export those external release
files.
Closes#6057
PIDs are pretty much always a bad idea because they can be racy. The
authorization did use the PID though. We can replace it by a check via
the system bus name.
Closes: https://github.com/flatpak/flatpak/issues/6212
For OCI remotes, the existing sideload repository system doesn't
work: identity for OCI remotes is done by manifest digest (disguised
as a fake commit ID internally), instead of by ostree commit, so
we have no way of knowing whether a sideloaded image matches the
summary.
Allow specifying a new form of sideload repository with:
--sideload-repo=oci:<path>
The desired use case for this is preinstalling Flatpaks during OS
install, and for this, binding the entire repository to a single
collection ID is both inconvenient and not useful, so OCI sideload
repostories don't have a defined collection ID - they just apply to
all OCI remotes. (And, because of this, they are restricted to
the command line.)
The memory management of returning a VarRefInfoRef is tricky - it
points to data owned by the summary or the sideload repo. External
consumers were always retrieving a copy of the summary metadata,
so make the public function do that.
