Closed clue closed 2 years ago
We have 0.0% users on HHVM, how about we drop support for it? https://packagist.org/packages/react/socket/php-stats#all
We have 0.0% users on HHVM, how about we drop support for it? https://packagist.org/packages/react/socket/php-stats#all
@WyriHaximus Dropping legacy HHVM is definitely an option as suggested in my OP. However, I would argue that dropping legacy HHVM actually requires more work at this point to avoid any unneeded leftovers. Given this changeset contains only a single line, my vote would be merge this as is and then look into this in a concentrated effort in the (possibly not too far) future 👍
look into this in a concentrated effort in the (possibly not too far) future 👍
@clue 👍
This changeset fixes the legacy HHVM build by downgrading Composer. Composer 2.3 does not longer support legacy PHP or legacy HHVM, so we have to downgrade to Composer LTS in order to run our test suite.
I'm filing this as an RFC to see if we want to apply this change to all our components (all HHVM builds currently fail). I'm not going to argue there's a lot of value in supporting a legacy HHVM build that accounts for less than 0.1% of our installation base, but at this point fixing legacy support is actually easier than dropping it entirely. At the same time, this is not the first fix for HHVM: https://github.com/reactphp/socket/pulls?q=hhvm
As much as I dislike having to address this, my vote would probably be to merge this as is and look into dropping legacy environments in a concentrated effort not too far in the future. WDYT?