Conversation
|
If it is "API and ABI compliant with 1.1.0", why do we need the aliases? Are there actual cases where 1.1.1 causes a failure? |
|
Not that I know of, but didn't want to break a convention by removing the |
|
Oh I just noticed that you actually did remove the The commit message is a bit confusing though. You don't actually update openssl, you just update the dependency for a few versions right? Or am I misreading that? |
|
I updated OpenSSL from 1.1.0 to 1.1.1. However, |
|
LGTM assuming there are no build failures @GrahamcOfBorg build ldns nodejs-10_x tor |
|
Failure on x86_64-linux (full log) Attempted: ldns, nodejs-10_x, tor Partial log (click to expand)
|
|
Failure on aarch64-linux (full log) Attempted: ldns, nodejs-10_x, tor Partial log (click to expand)
|
|
Timed out, unknown build status on x86_64-darwin (full log) Attempted: nodejs-10_x The following builds were skipped because they don't evaluate on x86_64-darwin: ldns, tor Partial log (click to expand)
|
pkgs/top-level/all-packages.nix
Outdated
| }) | ||
| openssl_1_0_2 | ||
| openssl_1_1_0; | ||
| openssl_1_1_1; |
There was a problem hiding this comment.
I think we should not introduce openssl_1_1_1 attribute, we should rename openssl_1_1_0 to openssl_1_1 and use that.
There was a problem hiding this comment.
No, it does not have to be changed in this PR, and it may be left as is: AFAIK openssl_1_0_1 and openssl_1_0_2 were incompatible.
There was a problem hiding this comment.
Please don't introduce openssl_1_0 alias for openssl_1_0_2 yet.
There was a problem hiding this comment.
Yes, please use openssl_1_1 and openssl_1_0_2.
There was a problem hiding this comment.
I don't feel strongly either way, but if future 1.1.x versions are expected to break backwards compatibility to 1.1.1, having that version explicitly makes sense to me.
| @@ -1,4 +1,4 @@ | |||
| { stdenv, lib, rustPlatform, fetchFromGitHub, pkgconfig, file, perl, curl, cmake, openssl_1_1_0, libssh2, libgit2, libzip, Security }: | |||
| { stdenv, lib, rustPlatform, fetchFromGitHub, pkgconfig, file, perl, curl, cmake, openssl_1_1, libssh2, libgit2, libzip, Security }: | |||
There was a problem hiding this comment.
Could you use openssl here, and call powerline-rs with openssl = openssl_1_1 in all-packages.nix?
|
Other than the change to |
|
The ofBorg failure is probably unrelated. It would be nice to still fix them in a separate commit though. They look like sandbox issues ( |
There was a problem hiding this comment.
Please replace this with openssl too.
88c72be to
9cef3ee
Compare
|
@GrahamcOfBorg build ldns nodejs-10_x tor powerline-rs |
|
Failure on aarch64-linux (full log) Attempted: ldns, nodejs-10_x, tor, powerline-rs Partial log (click to expand)
|
|
Failure on x86_64-linux (full log) Attempted: ldns, nodejs-10_x, tor, powerline-rs Partial log (click to expand)
|
|
Timed out, unknown build status on x86_64-darwin (full log) Attempted: nodejs-10_x, powerline-rs The following builds were skipped because they don't evaluate on x86_64-darwin: ldns, tor Partial log (click to expand)
|
|
nodejs-10_x builds locally for me. Seems to be a timeout issue but doesn't make much sense according to the log. @grahamc Is there a timeout per package? |
|
No, the timeout is per build including all dependencies. |
|
Then the question is how following error can be possible (from the x86_64-linux log): |
|
Hm that seems odd for a timeout. |
|
No attempt on x86_64-darwin (full log) The following builds were skipped because they don't evaluate on x86_64-darwin: tor Partial log (click to expand)
|
|
Success on aarch64-linux (full log) Attempted: tor Partial log (click to expand)
|
|
Success on x86_64-linux (full log) Attempted: tor Partial log (click to expand)
|
|
@GrahamcOfBorg build openssl_1_1 |
|
Success on aarch64-linux (full log) Attempted: openssl_1_1 Partial log (click to expand)
|
|
Success on x86_64-darwin (full log) Attempted: openssl_1_1 Partial log (click to expand)
|
|
Success on x86_64-linux (full log) Attempted: openssl_1_1 Partial log (click to expand)
|
Motivation for this change
From https://www.openssl.org/blog/blog/2018/09/11/release111/:
Added
openssl_1_0andopenssl_1_1aliases because that's the level of granularity packages almost certainly want, and will make future upgrades like this much easier.Things done
sandboxinnix.confon non-NixOS)nix-shell -p nox --run "nox-review wip"./result/bin/)nix path-info -Sbefore and after)