Don't try to muddle through, if the files expected to be present can't
be copied it's a hard failure.
Further abort with error code for failed commands so it can be caught by
the CI.
Signed-off-by: Ralph Sennhauser <ralph.sennhauser@gmail.com>
If running premake fails during update workspace then update workspace
needs to exit with an error as well so the CI will actually report an
error.
According to Itms testing with powershell passing the original error is
unreliable, so just exit with one.
Fixes: #7642
Signed-off-by: Ralph Sennhauser <ralph.sennhauser@gmail.com>
Shrinking GCs dump the JITted code, which leads to redundant recompilations, lowers performance, and makes profiling JS more difficult.
They may still happen if the runtime is at risk of OOM.
The cache would take the value post-player-modification, which is incorrect. This would usually just make caching fail, but could trigger worse problems if the post-player value is also used in other situations.
One such situation leads to an OOS for britons.
Thanks to elexis for the thorough investigation.
Fixes#7634
With the introduction of multiplayer loading screen tips, categories of
tips and probabilities of how often to show tips from a certain category
got introduced. The original probabilities ended up showing multiplayer
tips in 60% of the cases. As the number of multiplayer tips is still
fairly small, this resulted in very repetitive occurrence of the same
multiplayer tips over and over when loading multiplayer matches.
This commit changes that to give single-player tips a higher probability
for multiplayer matches. It also adds showing tips from the "Beginner"
category for multiplayer games, as there are regularly new players who
start their experience with 0 A.D. by playing multiplayer matches with
friends, instead of playing single-player matches and who'll therefore
benefit from those tips being shown for multiplayer matches as well.
New small metal mine models and actors.
New sparkle particles.
Remove refs to old metal ore mines from existing maps.
Fix refs to temperate_small_01_01, which should have been temperate_small_01
According to our stats `GL_ARB_fragment_program_shadow` is supported
by all our GL users. Also we're going to remove OpenGL ARB in A29. So
I remove ARBShadersShadow completely.
A multi player loading screen will have possibility of showing single
and multi player tips while multi player tips are in favor.
Tips are now separated by categories. During loading those category
have an occurrence probability, where we are choosing a random category
based on chance and not by complete randomness. From there we just
picking a random tip from that category.
This fixes the pulling of objects from LFS in the checkrefs workflow. To
do so three steps were necessary:
- Don't fail immediately when encountering missing objects in LFS
- Fix the remote URL of the fork repository
- Fix the authentication for the LFS storage of the fork repository
The option was missing, so player placement was not shown in the game
description part of the game setup, and it was not shown in the objectives
page in-game.
SpiderMonkey 98 introduced a size heuristic for the nursery GC region
(https://phabricator.services.mozilla.com/D136637). As this heuristic
uses a wall-clock time duration, it results in a severe performance
regression on slower systems for our use case.
This commit adds a workaround to turn off that heuristic, by telling
SpiderMonkey that a "page load" (something which doesn't have a meaning
in the context of pyrogenesis) is in progress, as that heuristic is
disabled for page loads.
Co-Authored by: Ralph Sennhauser <ralph.sennhauser@gmail.com>
Fixes#7714
Since 0eed117e6d exceptions from the map generation script are rethrown
in the main thread but not all of them are catched.
This defect is already fixed in the engine with 82513c9104Fixes: #7620
Profiler2 currently explicitly considers 'frame' for frames, which is a bit annoying in replays as state hashes are computed outside of these. It's better to just rely on RecordFrameStart() calls.
Up to now the chat autocompletion did complete words, even if there were
multiple candidates which matched available. So if for example the
autocomplete list contained "bar" and "baz" and a user typed "ba",
autocompletion would complete it to "bar", ignoring that "baz" was
available as possibly desired result as well. In practice this lead to
people regularly getting completions they didn't intend to get. This
commit fixes that by only completing text, if there is only a single
candidate available.
After 78900842b1, the following message was shown to clients when the
host disconnected during game setup:
"Lost connection to the server.
[Invalid value undefined]"
This patch fixes that.
The event property github.event.push.before seems to not be supported
and therefore the workflow as is fails on main.
Ignoring the trigger means a slight loss in coverage but as it still
fully checks PRs we are mostly good.
Signed-off-by: Ralph Sennhauser <ralph.sennhauser@gmail.com>
Add a workflow for linting and add a cppcheck job.
The job is using cppcheck provided by package manager. Existing errors
are suppressed to get going with CI linting, ideally tho the suppression
list would be empty. Some of the suppressed errors are only found with
more recent cppcheck than what is available in the runner.
Also remove old arclint cppcheck setup.
Signed-off-by: Ralph Sennhauser <ralph.sennhauser@gmail.com>
A shell utilities function library starting with a first utility
function for getting the number of online CPUs.
Signed-off-by: Ralph Sennhauser <ralph.sennhauser@gmail.com>
Theere is now an `Engine.startAtlas` property that will start Atlas
when it's returned from the root page. The `Engine.RestartInAtlas`
function is removed.
There are a few version of boost which deprecated global placeholders
but do not yet support std placeholders. Simply suppress warning for
those cases.
Signed-off-by: Ralph Sennhauser <ralph.sennhauser@gmail.com>
Use the single Debian 12 container image and inject CC, CXX and LDFLAGS
instead.
Further do the lfs checkout on the agent instead of inside the container
avoiding full rebuilds due to version mismatch.
Then split the debug and release builds using separate workspaces and
record warnings by build preventing gcc warnings leaking into clang
warnings.
Based on zero warning policy and incremental builds using a reference
build doesn't make sense. Removing that we can also drop an additional
generic workspace.
Signed-off-by: Ralph Sennhauser <ralph.sennhauser@gmail.com>
Debian 12 (bookworm) will soon be oldstable.
Most importantly this version comes with cmake-3.25, which will allow us
to use SDL3, cpp-httplib, fcollada fork and possibly others in the
future.
This also comes with gcc-12 and clang-14 by default, both implementing
almost all of c++17 and most of c++20.
Signed-off-by: Ralph Sennhauser <ralph.sennhauser@gmail.com>
Ralph Sennhauser