| Age | Commit message (Collapse) | Author |
|---|
|
|
|
Otherwise new one time keys will be uploaded on every start
|
|
|
|
|
|
|
|
|
|
On other platforms Docker fails to stand up the container.
|
|
|
|
|
|
It would probably be even better to pass the homeserver address in
the environment but that's a bigger endeavour.
Also: reformatted CREATE_CONNECTION macro.
|
|
|
|
This both is more reliable (GHA executes scripts in fail-fast mode) and
ensures that the return value is that of ctest.
|
|
QEventLoop refuses to work without an application object instance.
|
|
To use this in CI required extending/fixing autotests/run-tests.sh:
it can now accept arguments that are further passed to ctest invocation,
and it no more cd's to the build directory because build directories
can be in all kinds of places, expecting the caller to pick
the directory upfront.
|
|
|
|
QKeyValueIterator::operator->() only arrived in Qt 5.15.
|
|
To ensure Q_DECL_EXPORT/Q_DECL_IMPORT macros are defined.
|
|
The whole cycle will get terribly long again; looking forward to
parallel quotest...
|
|
For EncryptedFile:
- JSON converter bodies moved away to .cpp;
- instead of C-style casts, reinterpret_cast is used to convert from
(const) char* to (const) unsigned char*;
- the size for the target plain text takes into account the case where
the cipher block size can be larger than 1 (after reading
https://www.openssl.org/docs/man1.1.1/man3/EVP_DecryptUpdate.html).
- file decryption is wrapped in #ifdef Quotient_E2EE_ENABLED, to avoid
OpenSSL linking errors when compiling without E2EE.
|
|
That's a better primitive than just exposing SupportedAlgorithms list.
|
|
We don't seem to need sorted associative containers in those cases.
|
|
|
|
Using a static variable is incorrect as it doesn't load the device list
for any subsequent created Connection object.
|
|
A note on switching to QLatin1String for JSON key constants - this is
more concise and barely affects (if at all) runtime performance (padding
each QChar with zeros is trivial for assignment; and comparison can be
done directly with the same performance as for two QStrings).
|
|
Also, removed Q_GADGET macros from key verification events as those
don't seem to do anything (no Q_ENUM/Q_FLAG things, namely).
|
|
Don't set lifetime as version in call invites
|
|
|
|
This reverts commit 2cf44607cf0f057e147c2c4fe6dded6c13c58a8a (that was
stupid, honestly).
|
|
Logging categories used by Quotient are not supposed to be exposed
externally, which basically forbids usage of logging in header files.
A more flexible solution would involve moving logging.h to private
headers but Quotient doesn't have that thing yet.
|
|
The result is FTBFS as yet; next commits will fix that, along with a few
other things.
|
|
|
|
|
|
Previously you could not use toJson() on a polymorphic structure such
as one of those defined in eventcontent.h because it is not assignable
and the default specialisation of JsonObjectConverter used assignment.
To avoid that limitation, one had to specialise JsonObjectConverter for
each descendant of EventContent::Base, which is a lot of boilerplate.
The new JsonConverter (the template underpinning the "default"
to/fromJson implementation) improves on two things:
1. dump() allows to construct your own QJsonObject - or anything else
convertable to QJsonValue - instead of requiring to fill in the
pre-constructed one.
2. load() allows to construct your value type directly from QJsonObject
instead of default-constructing it in advance.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Co-authored-by: Alexey Rusakov <Kitsune-Ral@users.sf.net>
|
|
|
|
|
|
Co-authored-by: Alexey Rusakov <Kitsune-Ral@users.sf.net>
|
|
Co-authored-by: Alexey Rusakov <Kitsune-Ral@users.sf.net>
|
|
|
|
|
|
Co-authored-by: Alexey Rusakov <Kitsune-Ral@users.sf.net>
|
|
SyncData can load room objects out-of-line. This is only expected when
loading data from the cache (and since quite long ago, the cache always
saves room objects out of line, avoiding too large JSON payloads that
Qt parser chokes on). However, the code processed /sync response in
the same way; in particular, this meant that SyncData filled the vector
of unresolved room ids even when it came from /sync. SyncJob then looked
at this vector and entered an error state if it was not empty. Well,
payloads from the wire can be weird and it ultimately came to pass that
a homeserver returned a non-object against a given room key, triggering
the unresolved rooms branch in SyncJob - and stalling the whole sync
loop as a result (https://invent.kde.org/network/neochat/-/issues/500).
With this commit SyncData only fills unresolvedRoomIds when loading
rooms from the cache (with the implied fallback of discarding the cache
and loading from /sync anew instead). Respectively, SyncJob must never
end up with SyncData that has unresolved rooms (even if those occur
in the actual payload like in the mentioned issue, those rooms will be
completely empty instead); the added assertion only guards for internal
consistency.
|
|
Move out current room state to its own class
|
|
Because Apple Clang choked on `explicit(false)`.
|
