Ports on Big Sur x86 - portmidi

Ryan Schmidt ryandesign at macports.org
Fri Jan 15 17:12:12 UTC 2021


On Jan 12, 2021, at 10:51, Christopher Nielsen wrote:

> A few of our ports are failing on the build infrastructure, for target Big Sur x86_64. And one of the culprits is portmidi, which causes downstream ports to fail during the install-dependencies phase because "Dependency 'portmidi' with variants '' has previously failed and is required.”
> 
> But if you look at the build history for portmidi, there have only been two attempts for Big Sur x86_64: Details for the first (11/20/2020) don’t exist, presumably because they’ve been purged by now. The second (12/14/2020) shows failures at a higher level, but when you delve into the details, logs are missing:
> 
> https://build.macports.org/builders/ports-11_x86_64-builder/builds/7287
> 
> Regardless, if you scrutinize the ‘install-port’ step, it failed mighty quickly… in just 13 seconds. Without logs it’s hard to say - perhaps this is a legitimate failing during configure? - but still suspicious.
> 
> Has anyone with Big Sur on Intel, been able to successfully install portmidi?

Looks like Josh fixed this after your email:

https://github.com/macports/macports-ports/commit/51ce8ffec65af2eb0b430557f6005f2c114de092

We now have arm64 and x86_64 Darwin 20 packages for portmidi built on January 12:

https://packages.macports.org/portmidi/

And the build information is shown at:

https://ports.macports.org/port/portmidi/builds

FYI if you're looking through the build history on the ports.macports.org web page, old history will only show you builds where that port was directly built. It will not show you builds where that port was indirectly built as a dependency of another port. This deficiency in our build infrastructure was fixed earlier this month by Josh so the history will be more complete from here on out. https://trac.macports.org/ticket/60935

Also note that even though portmidi is now fixed on Darwin 20, any ports that had failed because of a portmidi dependency before will not automagically get tried again. It would be nice if we enhanced the build infrastructure to do that, but we haven't. If there were any specific ports you were thinking of that had failed due to portmidi, we can force builds of those, or if they need to be updated or changed anyway, you can change them and they'll get built in response to that change.



More information about the macports-dev mailing list