IF YOU WOULD LIKE TO GET AN ACCOUNT, please write an
email to Administrator. User accounts are meant only to access repo
and report issues and/or generate pull requests.
This is a purpose-specific Git hosting for
BaseALT
projects. Thank you for your understanding!
Только зарегистрированные пользователи имеют доступ к сервису!
Для получения аккаунта, обратитесь к администратору.
Change our existing "view as [u8] API", and also add one
that does a view as `OsStr`. The motivation for the latter
is I noticed ithat `OsStr::from_bytes()` *doesn't* copy,
or rather it just copies the pointer value. Rust's lifetime
inference ensures that the returned lifetime matches the input array.
I think the previous code in `treefile.rs` was confused about this.
Closes: #1688
Approved by: jlebon
I was going to add another usage of this function, and I think the
gerror stuff is unnecessary - if we are handed a bad file descriptor
(or a fd pointing to a regular file) that's something where we should
just abort.
While we're here, I'd like to codify expected usage in the function
names here. If you like this I'll e.g. also change `str_from_nullable`
to `ffi_view_nullable_str`.
Closes: #1685
Approved by: jlebon
Minor regression from #1587. There were places that were still doing
`dnf_context_set_cache_age()` manually, but those calls didn't exactly
have the intended effect since the core now handled caching itself.
The actual result was that the metadata was still being updated, but not
during the `dnf_repo_check` pass that the core does, but rather the
`Importing rpm-md` pass it does right after. So then, we were
incorrectly printing `(cached)` even though we'd update it afterwards.
Switch to the new way of doing things.
Closes: #1686
Approved by: cgwalters
Drop the `force_refresh` boolean parameter since we only ever call it
with `TRUE`. I think this dates from an earlier implementation where we
did call it with differing values.
Small prep for next patch.
Closes: #1686
Approved by: cgwalters
Let's make use of the GitHub release feature to make it more prominent
on the "Releases" tab, but more importantly so that we can attach
vendored tarballs for downstreams. E.g. this will allow us to have a
correct `Source0` field in the Fedora spec file.
Related: #1683Closes: #1684
Approved by: rfairley
This is relatively uncontroversial functionality that has already proved
useful when helping folks debug their stuff. Let's promote it to the
stable interface.
Closes: #1682
Approved by: rfairley
It said so right there; "nuke in next version". No need to carry this
stuff forever. It's the final phase of feature promotion, like when one
officially moves out of their parents'.
Closes: #1682
Approved by: rfairley
In a lot of places we're abusing `io::Error(io::ErrorKind::InvalidInput)`
which is both verbose and inaccurate really. Maybe in some
places we should be defining custom errors, but eh.
I like the `failure` crate. Use it in just `utils.rs` for now.
Tweak our error handling FFI wrappers to accept `Display` since
all we do is convert the error to a string.
Closes: #1675
Approved by: lucab
To make it more obvious what the difference between "Importing metadata"
and "Importing" is, add "rpm-md" to the first and "packages" to the
second.
Closes: #1681
Approved by: cgwalters
While we had an override for both the Fedora `glibc-all-langpacks.posttrans`
version and the RHEL7-era `glibc-common.post`, there were two
problems.
First, the RHEL7 version's lua calls `rpm.expand()` internally
rather than flagging the script itself for expansion.
Second, we also need to disable rofiles-fuse for it.
Closes: #1678
Approved by: jlebon
Otherwise, the object might still own an idle source on the main
context, which will cause issues if another pull operation happens
again. This wasn't causing issues before because in places where we did
do multiple pull operations, we would reuse the same
`OstreeAsyncProgress` object, and the second pull operation *did* do
`ostree_async_progress_finish()`. But that's no longer the case now with
66761916.
Closes: #1676
Approved by: cgwalters
This is not strictly necessary since the progress is considered ended on
the client side when the transaction is finished, but let's be nice.
Closes: #1676
Approved by: cgwalters
We should prefer the smoketested ref in general since it's more stable.
There is an updated glib2 there now. I also tested that layering-relabel
works (at least locally).
Closes: #1676
Approved by: cgwalters
In general our current CI/test system is susceptible to drift
between the container and AH. The direction we should be
going is to have coreos-assembler solve this problem with
a SDK, but for now, let's ensure that the container's libsolv
makes it to the host, same thing we do for libostree.
Closes: #1676
Approved by: cgwalters
This turned out to be messier than I thought, because of two primary
factors; the biggest mess here of course is the indirection
through the DBus API.
The other problem is that previously we passed the string to render
each time, and with current indicatif that'd trigger a rerender.
Since (usually) don't change the "prefix string", rework the API.
Change the "percent/n_items" bits to use autocleanups as well, and
to take the prefix string as an initial argument.
Since the state expands to multiple components, also change the
API to use the `0-initialized` pattern rather than trying to
return an aggregate.
We also gain a "sub message" which we use to display e.g.
package names as we're doing checkouts. Note this ends up
at the end, since otherwise everything else jumps around.
Closes: #1661
Approved by: rfairley
Prep for indicatif, the new progress implementation, which is now
more strict about overwriting tasks. The `OstreeAsyncProgress`
object lingered on and could own tasks on the mainloop. Narrow
the scope and avoid having one that crosses multiple pull requests.
Closes: #1661
Approved by: rfairley
The previous behaviour was to simply return "Invalid signature"
if the corresponding GPG public key wasn't found.
This status message wasn't clear enough that the key is missing.
If the GPG public key is now missing, a corresponding status message will be issued.
Closes: #1650
Approved by: rfairley
Otherwise we can get fun undefined behaviour like the caller thinking
there was a change when we didn't even install anything.
Closes: #1669
Approved by: cgwalters
That was the last bit in which we referred to the old unified core
option. Also change it in the source itself for completeness.
Closes: #1668
Approved by: cgwalters
We don't need to run the bwrap self test if we just want to print the
manifest. I played with putting the self test in `impl_install`, though
e.g. `postprocess` also needs this so it wasn't quite right.
Closes: #1666
Approved by: cgwalters
Previously, we were limiting the target repo in unified mode to be a
bare-user repo located on the same filesystem (see message of previous
commit). This patch lifts this restriction by making a distinction
between the *build repo* and the *final* target repo.
To do this, we create a bare-user repo located near the pkgcache to
take advantage of hardlinks and devino caching at commit time. And only
after committing do we essentially `pull-local` into the final target
repo. This of course allows us to avoid potentially pulling across the
two filesystems file objects that are already present in the target
repo.
This will be used by coreos-assembler:
https://github.com/coreos/coreos-assembler/pull/190Closes: #1490Closes: #1657
Approved by: cgwalters
This ensures that we always get hardlinks when checking out of the
pkgcache. This works right now because we indirectly require the target
bare-user repo and the pkgcache to be on the same filesystem by setting
`no_copy_fallback` in the core (I say "indirectly" because that setting
only enforces the workdir to be on the same filesystem as the pkgcache
repo, but since the workdir is currently placed inside the bare-user
repo...).
However, I'd like to change the requirement of a bare-user repo so that
one can commit into a repo on a different file system or a repo of a
different type (e.g. archive repo). This is prep for that.
Closes: #1657
Approved by: cgwalters
Perhaps an unexpected side benefit of slow compilation processes
is that one has an opportunity to reflect and ponder.
I realized during exactly such a moment that since we moved
`cbindgen` out of our library build, there's no need to wait
for the library to be built before we can start building the C
code.
This is a notable local quality-of-life development improvement.
Closes: #1665
Approved by: jlebon
Rather than defaulting to the host system's SELinux policy, we can be
much more efficient here if we instead use the policy of the last commit
if available. Likely, the pkgcache is currently labeled with that
policy, which means we skip the relabeling phase before checkout. But
also, if the policy didn't change at all in the new rootfs, we also skip
the second relabeling phase after assembly.
Closes: #1659
Approved by: cgwalters
External tools often want to parse the ref; for example coreos-assembler
currently does so. Let's ensure `${basearch}` is expanded with
`--print-only` so they can parse that JSON to get the expanded version
reliably.
Implementation note: this is the first Rust code which exposes a
"GLib-like" C API, notably with GHashTable, so we're making more use
of the glib-rs bindings.
Closes: #1653Closes: #1655
Approved by: jlebon
Make sure we always pass an integer and not a float to
`glnx_console_progress_text_percent`. And also make sure we clamp the
value we pass to 100. Not sure how I hit this, somehow the byte counting
we do upfront before committing is *less* than the addition of all the
`g_file_info_get_size()`s?
Closes: #1658
Approved by: cgwalters
Didn't hit this but this will be useful for a following commit which
delays committing to the final repo until the end.
Closes: #1658
Approved by: cgwalters
We don't actually need the repo in `rpmostree_passwd_compose_prep()`
since we're already passed the OstreeRepoFile root of the previous
commit as a separate arg.
Closes: #1658
Approved by: cgwalters
Just a minor readability tweak. Wasn't clear what `read_xattrs_cb`
meant. Also inline an indexing var into the loop.
Closes: #1658
Approved by: cgwalters
If the policy didn't change, then it's safe to use the devino cache
regardless of the repo type. This dates from #1123; I think the
rationale there was so that it only affects the pkglayering path, where
the system repo is bare. But really, it should be fine to generalize
this since we don't have any xattr callbacks.
Closes: #1658
Approved by: cgwalters
This is slight bikeshed, though this comment took me a second to parse
because it says "not do it" right before we do it. Let's just move it up
to right outside where the condition is actually tested.
Closes: #1658
Approved by: cgwalters
No reason to do this sanity check so late. Let's bring it up to near the
top of `impl_install_tree()` so we avoid useless work if possible.
Closes: #1658
Approved by: cgwalters
We don't actually need this helper here since we always initialize
`self->cachedir_dfd` nowadays. I took a closer look at this initially
because it was confusing ctags due to the function being named the same
as the struct member it returns.
Closes: #1658
Approved by: cgwalters
In #1630, we lowered SELinux policy loading into the core. However, this
also enabled SELinux policy loading from the host system even in the
legacy (non-unified) compose path. This meant that compose systems now
needed to have the policy installed even though we didn't need it at
all. This caused regressions in pungi:
https://pagure.io/dusty/failed-composes/issue/956
Just make the binding of the "selinux" member conditional on whether or
not we're in unified mode (which is really when we even care about
having it loaded from the start for pkgcache purposes).
Closes: #1656
Approved by: cgwalters
Wrap the `Version` key in the YAML-compatible output of
`rpm-ostree --version` with quotes so that it's parsed as a string. The
issues with the previous approach in a nutshell:
```
In [5]: yaml.load("asdf: 2018.10")
Out[5]: {'asdf': 2018.1}
```
It's treating the version number as a floating-point. Now, this is
technically a backwards incompatible change, but given that the previous
approach is inherently broken for our needs, I don't see a way around
breaking it now.
See also: 759b099861Closes: #1654
Approved by: cgwalters
There is no `rpm-ostreed-automatic` man page, only
`rpm-ostreed-automatic.service` and `rpm-ostreed-automatic.timer`. Fix
references from the conf man page to those pages.
Closes: #1651Closes: #1652
Approved by: miabbott
This currently requires a `--i-know-this-is-experimental` flag;
I know it'd be a bit more consistent to have it under `ex`, but
what feels weird about that is *most* of the `ex` commands people
use are client side. This is where we want it to ultimately end
up.
We've landed a lot of prep patches, but I know there's still
a notable amount of code duplication with `compose tree`. What's
left is about ~700 lines but it's mostly not hard/complex code
anymore.
In the future, I'd like to extract more of the compose code
to a `rust/src/compose.rs` or so, but I think this is sustainable
fow now.
My high level goal is to get this into coreos-assembler and stand
up a Silverblue build that uses it.
Closes: #1512
Approved by: jlebon
