extra-php-extensions
extra-php-extensions copied to clipboard
brefphp
Support ARM64 versions now Bref 2 has landed
I'm no expert when it comes to docker and compiling these layers. What would be involved in shipping layers that support ARM64?
The key one that's blocking us from moving to ARM/Graviton is Relay/Redis.
I think the short term solution if you really want the ARM upgrade to save money is to build the layers you need in ARM using docker buildx and create a private AWS layer in your account.
Just use the existing Dockerfile code available in this repo. Double check if any of the code would crash in ARM architecture.
I'm going to wait.
Hi both,
To answer the original question, what needs to be done is to finish https://github.com/brefphp/aws-lambda-layers/issues/62
Then we can work on updating the build here (it's quite a bit of work to be honest, not a matter of a few hours)
it's impossible/hard to build multi-platform images
I might say something really stupid but why are you trying to build a multi arch image for plugins? Why not just ask devs to pick the right arch image? If I want redis ARM, why not just build an image that only supports ARM? Would that help?
Multi arch image sounds very convenient to have but the tooling is not really following it seems...
For the record I've continued looking into ARM today, here's an update: https://github.com/brefphp/aws-lambda-layers/issues/62#issuecomment-1555955615
TL/DR slow progress, it will take more time.
@mnapoli any chance this will be finished soon? In the meantime, is there an option for single-arch builds for at least most popular extensions? (e.g. gd?)
@chekalsky unfortunately I have made very little progress. Extensions are one of the most time-consuming things with Bref.
That's not something I want to do, but so far the only viable solution I see is to offer "official" extensions in the Bref project via a paid option (the bref-extra-extensions repo is community-maintained, I try to contribute when I can). That would allow to invest time to maintain the extensions properly. I'm not sure it would be a popular move though, so that's why I haven't done it yet.
I understand. With reasonable pricing we would love to contribute, but for now it is not that big of a difference for us in migrating to graviton to offset full-scale contribution/support for that feature. Hope someone bigger could invest in that, or you would implement payments.