The Coverity scan app can't handle GCC 11 yet, and will complain about
things inside the stdlib headers and only compile 15 files. So we need
to switch back to Fedora 33 and use GCC 10 instead.
* Try bumping Coverity up now that we should be on a newer version
of it (so hopefully it supports GCC versions with 2 digits in their
number).
* Allow the metadata test to fail because of an upstream bug in the
appdata validation program that thinks the description tag isn't
translated even though it is.
* Switch to using org.kicad.kicad as the main app ID
* Move the appstream file to a metainfo file
* Rename all metadata files to use the reverse DNS scheme
* Add header matching to mime type files where possible
* Add a mime type for gerber job files
* Add a QA test that verifies all the linux metadata files and
run it in CI
* Update the visible names in the launcher entries to be more
consistent with the internal visible names
* Update various strings and items in the metainfo file so they
are more descriptive and easier to read
* Don't run the Ubuntu job on schedules
* Fix Coverity job by being explicit it is for schedules only
* Separate the caches for the Ubuntu and Fedora jobs
Eventually we probably want to just pick one target for it,
but for now this means we can get faster test results as long
as we have a faster Ubuntu runner.
The original test didn't work if the tgz was not already in the
runner cache. This new test will fail the check if the file is missing
instead of erroring the command.
This moves the program-specific code (e.g. BIU files) into
the program tests.
Also, create title_block.cpp to break a dependency that pulled
in eda_text.cpp when using the TITLE_BLOCK object.
Seth Hillbrand
Ian McInerney
Roberto Fernandez Bautista
Mark Roszko
stefan_test
Jon Evans