| Age | Commit message (Collapse) | Author |
|---|
|
[skip ci]
|
|
|
|
:latest stopped working for some reason.
|
|
GCC 10 ICE's[1] in qt_connection_util.h code; and ubuntu-20.04 doesn't
have GCC 11.
Also: patch a Qt 5.15 header when compiling with GCC because a
combination of Qt 5.15 and GCC 11 in turn triggers QTBUG-91909/90568...
Which in turn required moving Qt setup before the build environment
setup. Life's fun.
[1] Internal Compiler Error
|
|
Apple Clang doesn't have those yet.
|
|
...thanks to C++20 awesomeness. A notable change is that
wrap_in_function() (and respectively function_traits<>::function_type)
and fn_return_t alias are gone. The former are no more needed because
connectUntil/connectSingleShot no more use std::function. The latter
has been relatively underused and with the optimisation of switchOnType
hereby, could be completely replaced with std::invoke_result_t.
Rewriting connect* functions using constexpr and auto parameters made
the implementation 30% more compact and much easier to understand
(though still with a couple of - now thoroughly commented - tricky
places). Dropping std::function<> from it may also bring some (quite
modest, likely) performance benefits.
|
|
|
|
Too many parameters of the same type in a row.
|
|
|
|
|
|
Since this object has to be verified against a signature it also carries
there's a rather specific procedure described in The Spec for that.
That procedure basically assumes handling the signed one-time key
object as a JSON object, not as a C++ object. And originally Quotient
E2EE code was exactly like that (obtaining the right QJsonObject from
the job result and handling it as specced) but then one enthusiastic
developer (me) decided it's better to use a proper C++ structure -
breaking the verification logic along the way. After a couple attempts
to fix it, here we are again: SignedOneTimeKey is a proper QJsonObject,
and even provides a method returning its JSON in the form prepared for
verification (according to the spec).
|
|
|
|
Now there's only 1 instead of 5 lookups of the same EncryptionEvent,
and the code is shorter.
|
|
In keyverificationevent.*, this massively shortens repetitive getter
definitions; the remaining few non-trivial ones are moved to
keyverificationevent.h, dropping the respective .cpp file and therefore
the dedicated translation unit. In roomkeyevent.h, it's just shorter.
|
|
|
|
toJson(SignedOneTimeKey) incorrectly generated a "signatures" key
mapped to an empty object when no signatures were in the C++ value.
Also: fallback keys have an additional flag that also has to be taken
into account when verifying signatures.
|
|
'performance-no-automatic-move' triggers on code where copy elision
normally takes place anyway. In fact, all cases it triggered on were
also subject to named return value optimisation (NRVO).
[skip ci]
|
|
Honestly, it was quite intuitive even without that, but in reality
there are implicit conversion under the wraps. This commit makes them
explicit, for clarity.
|
|
Also: leave a link at the place in the spec with power level defaults
to make it clear they are not invented out of thin air.
|
|
|
|
Also: make ImplPtr more flexible.
|
|
|
|
|
|
|
|
GCC (even 12.x) doesn't like when a template parameter is of
a pointer/reference type and dumps this warning. See also:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90670
|
|
|
|
...instead of tucking the template in filesourceinfo.cpp where it surely
will be forgotten.
|
|
|
|
|
|
|
|
|
|
This pertains to QUO_IMPLICIT and DECL_DEPRECATED_ENUMERATOR - both can
be used with no connection to Qt meta-type system (which is what
quotient_common.h is for).
|
|
make_array() has been introduced to cover for shortcomings on macOS and
Windows. These shortcomings are no more there, so we can just use the
standardrlibrary.
|
|
|
|
|
|
That `std::move(_data)` never worked because the passed object is
a precursor to RequestData, and RequestData always takes things by
const-ref or by value, never by rvalue. Also, explicit mention of
RequestData is unnecessary, as its constructors are implicit by design.
|
|
Also: add a Matrix chat badge.
[skip ci]
|
|
|
|
|
|
Also: build with Qt 6 first, so that it fails sooner.
|
|
Although Qt 5 didn't complain about that, you could never really use
sendToDevices() in its slot (or even invocable) capacity because
Qt's meta-type system could not handle move-only UsersToDevicesToEvents.
Qt 6 is more stringent; the build fails at trying to instantiate
QMetaType for that type (with a rather unhelpful error message thrown
by Clang, and more helpful but very verbose diagnostic from MSVC)
because it does not provide a copy constructor.
However, sendToDevice doesn't really need to have full-blown events
in that parameter; just the content of the event is equally fine.
This commit does exactly that: replaces UsersToDevicesToEvents with
UsersToDevicesToContent that contains QJsonObject's instead of
EventPtr's. The code around is updated accordingly.
Also: factor out the key event JSON creation from
makeMessageEventForSessionKey() because it's the same JSON for each
target device; the function therefore is called encryptSessionKeyEvent()
now.
|
|
- CodeQL analysis was executed on every job that ran Clang, humping
the total execution time by 10+ minutes alone. Now it only runs on
a single job.
- libolm is no more compiled but installed from the repo, along with
libssl-dev; and both are installed in the same transaction as ninja
and valgrind, shaving out one apt transaction
- One more Windows job has been added to test building with Qt 6.3.1
on that OS.
- Qt version is pushed earlier in the job matrix, as it becomes more
significant than the compiler for a given platform.
|
|
Qt 6 builds are allowed to fail for now.
|
|
Also, use MSVC 2019 on Windows.
|
|
Its platform has been lagging behind most of the time; but more
importantly, the value from its analysis is almost non-existent, with
just one considerable issue being identified over the recent year if not
more. These days we have clang-tidy and Sonar that are much better at
static code analysis.
[skip ci]
|
|
[skip ci]
|
|
CONTRIBUTING.md got bitrotten in quite a few places.
[skip ci]
|
|
|
|
It takes well over an hour to build the whole lineup for now; while
the single right fix for that is making quotest capable of running in
parallel, a few GCC jobs can be safely dropped for now (and we'll see
if they should be brought back when parallel quotest unleashes the CI).
|
|
...meaning, clients have to compile in C++20 mode too from now.
|
