Hey guys, we recently started to use B2B Frontend Image Optimization:

U019JM2HE7N

Hey guys, we recently started to use B2B Frontend Image Optimization:
https://github.com/spryker-shop/b2b-demo-shop/blob/master/frontend/libs/images-optimization.js
Since these modules depend on additional dependencies like a C-Compiler (make, autoconf, …) we had to extend the current cli container setup.
https://github.com/spryker/docker-sdk/blob/master/images/common/cli/Dockerfile#L17
 
Is this on purpose not included out of the box or did we miss something here?

sprymiker

@UTJU5PD6F Is there a way to use image optimization without installing gcc into docker/sdk images?

Alex Malkevich

It's installation is first done by downloading precompiled sources from network and then fallback to compiling. So if the base image is not recognised you always need gcc to build it.
This is an example of the list of supported platforms https://github.com/imagemin/mozjpeg-bin/tree/master/vendor

sprymiker

@UTJU5PD6F So why does not it find precompiled sources? What’s a preconditions?

Alex Malkevich

Could be many reasons - incompatible binary or detection mechanism is incorrect in docker lightweigt images.
It uses some custom "BinWrapper" to determine which binary to use here https://github.com/imagemin/mozjpeg-bin/blob/ffd81a96301a6df3004c011c4f84b8a9fc63e620/lib/index.js#L8-L13 and then it tries to execute it to test here https://github.com/imagemin/mozjpeg-bin/blob/ffd81a96301a6df3004c011c4f84b8a9fc63e620/lib/install.js#L10-L12
And based on those 2 outcomes it decides to either use precompiled one or compile from sources.
I would suggest to try and install in docker and check install logs to understand what step fails.