2020-09-20
Members present: James S, Joshua Parmenter, patrick, brian, Marcin, elgiano
macOS11 problems (Joshua Parmenter):
- example:
SynthDef(\test, { Out.ar(0, SinOsc.ar(440, 0, 0.2)); }).add // works
s.sendBundle(nil, [\s_new, \test, 1000, 0, 1] ) // doesn't ever schedule
s.sendBundle(0, [\s_new, \test, 1001, 0, 1] )
most probably a problem with endiannes with ntohl.JP is working on it
-
Qt problems. Waiting for next Qt release for macOS (Qt currently needs to be built from source to target macOS 11).
-
Binary download don’t work on macOS 11. Compiling from source works. Need to announce this. JP is announcing on sc-users.
-
We need both x86 + arm releases (two separate builds) when we start to support macOS 11.
macOS CI error (Marcin)
- errors during testing. Seems related to no graphics env available → web engine failing.
- CI is blocked for macOS until this is fixed. Marcin is on it.
- Possible temporary workaround: disable tests.
3.11.2 / 3.12.0
- discussed in Issue #5146:
- Planning 3.11.2 for October.
3.12 right after:
- macOS 11 support if possible by then, but not if it would delay the release.
- macOS server-side volume clipping ?
Would there be a possibility than upgrading Qt could drop an older system we currently support?
- If it happens we could include a pre-processor switch in a patch release. (add a new build job to do that)
Talking and reading about the future of Qt.
James S: Qt 6 is not advertising breaking changes, but won’t work without C++17.
- Setup a Qt 6 working group, or issue, to keep track of it.
- Require C++17 (open ticket for C++17 in sc as well)
- Setup a C++17 working group, or issue.
Does C++17 allow new features for sclang?
- Not out of the box
- drop a lot of boost support?
- lock/mutexes in supernova
Maybe a C++17 ticket on github is enough? Let’s start with that.
Websockets (patrick)
(see RFC #9 for previous discussion points)
with or without Qt?
- without Qt would be better for headless builds
- better to avoid another dependency
Qt’s websockets are split from webengine and GUIs. So Qt could be viable for headless.
Commit notation (James S)
Conventional commits: standardized way of writing commit messages
-
machine and human readable
-
based on semantic versioning
-
tooling to automatic generate a changelog
-
might be asking too much to a community project
-
but it could be done by maintainers when PRs are merged
-
even if not 100% strict, would alleviate changelog-writing pain
IT IS PATRICK’S BIRTHDAY TODAY