You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I personally don't have a problem with that. Users are likely going to take note of the (lack of) performance of TLS and cryptography in general though.
Check the "supported platforms" section in BUILDING.md. You basically want to get riscv to (at least) tier 2, meaning there needs to be CI infrastructure and one or more people maintaining it (where "it" is both the infrastructure and the state of the test suite.)
That's to get to tier 2. To stay there is an ongoing commitment, and not of the "check in once a month" kind.
Missed these comments - thanks for closing this issue which is pretty much a dupliate and it's best to avoid fragmented comment threads, but to give closure on a couple of explicit questions above:
If openssl 3.1 is out of reach, could we accept that, for now, on RISC-V, openssl will be compiled without asm support?
That's what's happening in the unofficial builds currently. So it works, but likely not quick, although that may be the cause of some of the test case failures.
Do you know what infrastructure is used to build and publish the unofficial RISC-V binaries?
It's the same x64 host used for all of the linux unofficial builds so it's done with a GCC cross-compiler which is outside the noraml node.js CI as @bnoordhuis asserted. I don't do anything locally for them
As per @targos' comment, let's have any follow-on discussion in one of the other issues.
Activity
bnoordhuis commentedon Aug 26, 2023
No. The 3.1 branch is not LTS, like 3.0 is. From https://www.openssl.org/policies/releasestrat.html:
Node v20.x's EOL is in April 2026.
ilg-ul commentedon Aug 26, 2023
If openssl 3.1 is out of reach, could we accept that, for now, on RISC-V, openssl will be compiled without asm support?
bnoordhuis commentedon Aug 26, 2023
I personally don't have a problem with that. Users are likely going to take note of the (lack of) performance of TLS and cryptography in general though.
ilg-ul commentedon Aug 26, 2023
Right. But we can mention in the release notes that this is a temporary solution until RISC-V support in openssl will mature.
In this case, what would be the next steps to make this happen?
bnoordhuis commentedon Aug 26, 2023
Check the "supported platforms" section in BUILDING.md. You basically want to get riscv to (at least) tier 2, meaning there needs to be CI infrastructure and one or more people maintaining it (where "it" is both the infrastructure and the state of the test suite.)
That's to get to tier 2. To stay there is an ongoing commitment, and not of the "check in once a month" kind.
ilg-ul commentedon Aug 26, 2023
Thank you, that's a good starting point.
Do you know what infrastructure is used to build and publish the unofficial RISC-V binaries?
bnoordhuis commentedon Aug 26, 2023
I'm not 100% sure but I expect @sxa builds and uploads them locally, no CI involved.
bnoordhuis commentedon Sep 27, 2023
It's been a month with no movement. I'll take the liberty of closing this but LMK if there is reason to reopen.
targos commentedon Sep 27, 2023
Discussion is happening on #42251 and nodejs/build#2876
sxa commentedon Sep 27, 2023
Missed these comments - thanks for closing this issue which is pretty much a dupliate and it's best to avoid fragmented comment threads, but to give closure on a couple of explicit questions above:
That's what's happening in the unofficial builds currently. So it works, but likely not quick, although that may be the cause of some of the test case failures.
It's the same x64 host used for all of the linux unofficial builds so it's done with a GCC cross-compiler which is outside the noraml node.js CI as @bnoordhuis asserted. I don't do anything locally for them
As per @targos' comment, let's have any follow-on discussion in one of the other issues.