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.
This way we don't randomly pick up bits from the C library
unintentionally as things change on that side.
I think the support for `!` in `include` may be relatively new
and that's why the original author here chose to do things
via `exclude`. But using `include` with a few specific exclusions
is just way better.
We don't need them long-lived anymore. They were just used for reporting
at this point, but we can use the new `static_delta_targets` hash table
size for that now.
When we're fetching a commit via static delta, we already take care of
fetching the full commit, so there's no need to also scan it using the
regular object workflow.
Closes: #2053
If e.g. detached metadata or a parent commit is missing, we
don't consider it an error but we erroneously still increment the
`n_fetched_metadata` counter, causing it to be higher than it should.
This will implicitly be tested by a test added in the next patch.
- Also rolls up a few other changes in the C API
- Drop `&mut` from `OstreeKernelArgs`; it now confuses type inference.
More generally while using `&mut` here was well intentioned, it
goes against the glib standard of mapping everything to `&` and
accepting interior mutability.
Since it bumped semver (when I didn't expect it to; xref
963eebf3ab (r121651362)
It's not load-bearing enough here to matter versus just passing
an untyped file descriptor.
This mainly means that it will be the `glib` ecosystem which
forces transitive semver bumps for us, not both.
It complains that we could leak memory if the return value
pointer isn't set. That's actually a nonsensical case, there's
no reason to call this and ignore the return value.
So change things to require it be set, and also change the
`g_return_val_if_fail` to be hard assertions per our new policy.
We did this in some prior patches because it's better for
static analyzers; code is either reachable or not, and we don't
want the default desktop-style behavior of "try to stumble on
without crashing". It's just likely to lead to a crash somewhere
else.
The warning here was about a use-after-free which is serious, but
I don't think it was reachable in practice, because we'd only
enter the second conditional if `contents_checksum != self->contents_checksum`,
yet we made them the same thing here.
Nevertheless, refactor the code to do a direct return in this
case because that's basically what happened before, and doing
so is a lot clearer.
