A symbol from this (flatpak_portal_error_quark) was leaked to libflatpak
due to being marked FLATPAK_EXTERN, so to keep ABI we move it back.
Fixes https://github.com/flatpak/flatpak/issues/1613Closes: #1616
Approved by: alexlarsson
This means we track why an operation was queued. For example, that a runtime
installation/update was queued because an app needed it. We then mark operations
that failed, so that the app will not be installed if the runtime install failed.
There is one case where we do allow the installation to succeed, when an app
relies on a runtime which is installed, but the update of it failed. The
app should then still work, and we don't want to block installation/update
of an app if the runtime repository is down.
Closes: #1611
Approved by: alexlarsson
On 32 bit ARM platforms, flatpak_get_compat_arch() returns NULL, so
handle that gracefully in flatpak_get_arches().
Closes: #1614
Approved by: alexlarsson
The telegram app id is org.telegram.desktop, and its appstream
component id is org.telegram.desktop.desktop, which we did not
properly handle (we special cased the app-id-ends-with-desktop case
and then did not remote .desktop). This replaces that with a more
approach that *always* matches the whole app id as prefix, and then
replaces a ".desktop" in the suffix part only.
Closes: #1593
Approved by: alexlarsson
This works just like in build-finish, but the main difference is
that the extension point is added early, so that the build itself can use it.
Closes: #1598
Approved by: alexlarsson
Sometimes we're not able to access the `xa.cache` data that's usually in
the summary and ostree-metadata, such as when the remote is a USB drive
that had `ostree summary -u` run on it. In that case, we don't want
flatpak_remote_ref_new() to fail, because the FlatpakRemoteRef instance
might still be useful enough without the metadata, such as in
flatpak_installation_list_remote_refs_sync. So this commit makes the
metadata property of FlatpakRemoteRef nullable which has the effect of
making the test_list_refs_in_remote() test pass with P2P enabled.
The metadata property wasn't added to FlatpakRemoteRef until commit
c33f842b7 so presumably no one is depending on it.
Closes: #1600
Approved by: alexlarsson
This makes flatpak_remote_state_lookup_cache() return the metadata
from the actual FlatpakRemoteState rather than duplicating it.
All callers are updated to not free anymore and to strdup if needed.
Closes: #1594
Approved by: alexlarsson
This is safe as we do the pull locally, and root configured the local
path. We already do the same for regular installs, so its weird that
its not done for appstream.
Should fix https://github.com/flatpak/flatpak/issues/1580Closes: #1585
Approved by: alexlarsson
By default we use the new appstream2 branch if it exists in the remote,
also in this case we compress the xml when deploying to be backwards
compat with the old deploys.
Closes: #1585
Approved by: alexlarsson
For example, if a i386 build is in the repo but no x86-64 version then
also add the i386 build to the x86-64 appstream data. However, don't
add it if that would cause a duplicate (i.e. both the x86-64 and i386
version).
Closes: #1585
Approved by: alexlarsson
This early bailout is not really needed, because noop updates is
pretty fast. Also, doing that breaks the timestamp updates.
Closes: #1585
Approved by: alexlarsson
This fixes the ability of the remote-ls command to take a file:// URI
instead of a remote name, which is especially useful for repos on USB
drives (created via `ostree create-usb`) which are temporary and don't
warrant being added to the repo config. This commit also updates
relevant documentation, adds a unit test, and updates a few variable
names to improve readability.
I can't find a commit in the history where this was working, but it's
working on the Endless fork of flatpak so I think there was agreement at
some point that it's desired behavior.
Fixes https://github.com/flatpak/flatpak/issues/1588Closes: #1587
Approved by: mwleeds
Currently _flatpak_dir_get_remote_state() only fetches the
ostree-metadata ref if it was able to fetch the remote summary, but this
is unnecessary because we don't need to know the checksum just to fetch
it, and this is especially problematic in the offline use case when the
remote summary can't be fetched. So this commit makes flatpak fetch
ostree-metadata even if the summary fetch failed, which is consistent
with how things worked before commit fedb0e5bd.
Closes: #1587
Approved by: mwleeds
When getting the dynamic remotes for the installed refs, if no refs
with a collection ID are found, then the ostree_repo_find_remotes_async
is given an array with just NULL, which makes it early return and thus
it would result in an infinite loop in
flatpak_installation_list_installed_refs_for_update.
This patch only tries to find the dynamic remotes when there are such
refs, thus preventing the mentioned error and any extra unnecessary
processing.
Closes: #1587
Approved by: mwleeds
In list_remotes_for_configured_remote(), we call
ostree_repo_finder_avahi_start() without checking the error status. This
means the OstreeRepoFinderAvahi instance passed to
ostree_repo_find_remotes_async() could be in an error state, which leads
to flatpak getting infinitely stuck waiting for the result of that
operation. So this commit removes the OstreeRepoFinderAvahi instance
from the array if it failed to start, and returns early if only LAN
remotes were requested. This is also consistent with how ostree itself
handles an error starting the Avahi finder instance in case NULL is
passed to ostree_repo_find_remotes_async() instead of a finders array.
This error condition happens on the Endless OBS server because there's
no Avahi daemon in the chroot used to run "make check".
Closes: #1587
Approved by: mwleeds
Fetching a remote ref may fail because of the locally configured
collection ID not matching the remote's one (because it's never been set
in the configuration or it's just different/incorrect),
However, using `flatpak_installation_fetch_remote_ref_sync` there's no
mention of the collection ID to the caller, so we should make sure we
try to do our best to match the given parameters. Thus, if the lookup
fails (because the collection IDs don't match), we resort to iterating
over every ref and comparing their names with the expected ones.
This should be however reworked, through a new method likely, in order
to accept the collection ID as an input argument, so the caller has more
power to choose which one to match.
Closes: #1587
Approved by: mwleeds
When listing the removable remotes, it will stop the operation after the
first configurable remote that has no collection ID set.
This is of course a problem as such an occurrence is not a reason for
not trying to list any remaining remotes.
This patch fixes that by simply reporting the failures as a debug
message instead of aborting.
Closes: #1587
Approved by: mwleeds
When p2p is enabled, listing the remotes using
flatpak_installation_list_remotes will include all types of remotes.
However, this can lead to problems for the users since dynamic remotes
(USB/LAN) may not be used always the same way that static ones are. For
instance, when listing the refs in a remote, dyanmic ones cannot be
listed by name; so any existent code that iterates through the list of
remotes and lists the refs in them will not work properly.
To avoid this situation, this patch makes the API method in question to
return only static type remotes. Getting all remotes can still easily be
accomplished by calling flatpak_installation_list_remotes_by_type with a
0-sized array (or by specifying all types).
Closes: #1587
Approved by: mwleeds
Since we have several types of remotes, it is important to have the
option of choosing which types one wants when listing them because
depending on the type it may have an impact in the performance (e.g.
when listing LAN remotes).
For that reason this patch adds a new method
"flatpak_installation_list_remotes_by_type" that allows to specify
which types of remotes should be returned. When giving an empty array
of types, it means that all types should be returned instead, which can
be more useful than actually asserting or not doing anything.
Closes: #1587
Approved by: mwleeds
When listing refs from remotes, a remote can be given by name or by URI.
This is an important distinction because dynamic remotes (USB/LAN) are
not internally mapped to a name, and thus their generated name cannot be
used for listing them. Thus, the solution is to use their URI in order
to list the refs directly from it.
Even though this feature has been around forever, the documentation
didn't really reflect it, so this patch mentions the described
alternative possibility.
Closes: #1587
Approved by: mwleeds
This is in order to allow tests to make several apps if needed, and will
be useful when e.g. testing repos that have multiple apps with different
collection IDs.
This patch modifies the mentioned script as mentioned and updates the
places that call it.
Closes: #1587
Approved by: mwleeds
When listing refs installed or from a remote, only the refs matching the
main collection-id were being returned. However, it is very important to
have access to all refs, independently from their collection-id,
especially when trying to list remotes coming from a USB repository.
These changes add the mentioned refs and update the places that use
this list, both in the lib and in the CLI.
For the implementation to become easier, we introduce also a
FlatpakCollectionRef, that should be replaced by OstreeCollectionRef
once the latter becomes part of the general API (currently it is in the
experimental one).
Closes: #1587
Approved by: mwleeds
Since we can now get the URL of the remote directly from the
OstreeRemote object, we should use it when creating a FlatpakRemote.
This will allow temporary FlatpakRemote objects (corresponding to
removable mounts) to have a URL set to them.
Closes: #1587
Approved by: mwleeds
This adds things like the size and the metadata, as well as eol strings
to FlatpakRemoteRef. We typically have this accessible anyway, in the
FlatpakRemoteState.
This makes flatpak_installation_fetch_remote_size/metadata_sync deprecated.
Closes: #1591
Approved by: alexlarsson
To implement this, we have a concept of a custom export filter
function which can be specified for each path to determine the
files that can be exported under that path.
Closes: #1589
Approved by: alexlarsson
