fabric-ca-client and fabric-ca-server build fails on s390x in v1.5.7 - likely due to commit d575f40e about fabric-ca HOT 5 CLOSED

A recent github.com/IBM/mathlib update introduced an architecture specific build tag.

I see that a "generic" implementation for mul() is available here:
https://github.com/IBM/mathlib/blob/main/driver/kilic/custom_generic.go#L53

This file has a build tag of //go:build !amd64 || generic:
https://github.com/IBM/mathlib/blob/main/driver/kilic/custom_generic.go#L1

Based on that build constraint, I would expect architectures besides amd64 to use the generic implementation automatically. But maybe the "generic" build tag needs to be explicitly passed?

@ale-linux Any ideas? What was the intent for architectures besides amd64?

from fabric-ca.

Thanks to @ale-linux investigation, we found that updating github.com/IBM/mathlib to v0.0.3-0.20231011094432-44ee0eb539da resolves the problem. PR #387 opened.

Note that it can be tested on other platforms by doing:

env GOOS=linux GOARCH=s390x make clean fabric-ca-client
and
env GOOS=linux GOARCH=s390x make clean fabric-ca-server

from fabric-ca.

@silliman @asararatnakar @benjsmi Fix has been merged into fabric-ca main branch. Can you try a s390x build from main branch, and if it looks good, we can release v1.5.8 with the fix.

from fabric-ca.

@denyeart Thank you for the follow up Dave. We were able to build s390x from main branch.

from fabric-ca.

This builds fine now on s390x. I successfully ran the fabric-samples/test-network (with -ca argument) with the built fabric-ca OCI image. Thanks for the fix!

from fabric-ca.