-
-
Notifications
You must be signed in to change notification settings - Fork 419
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Allow use of OpenSSL 1.1.1 when building Pony #3156
Conversation
❓ Shall we add some test build that actually uses openssl v1.1.1? |
Looks like we could either
What do you think? Also besides adding a Dockerfile, and a line to the CircleCI config, is there anything else that needs to be done to wire up the new container image? |
Maybe we should just replace the current testing of |
IMO it would make sense to support Ubuntu LTS releases and the latest non-LTS, so I'd go with @srenatus' second option. |
Happy to add what's required here! I'm still not sure if the Dockerfile and the mention in the circleci config is sufficient, but I guess I'll find out. 😃 |
💡 https://github.com/ponylang/ponyc/blob/master/.ci-dockerfiles/ubuntu-18.04-base/README.md#push-to-dockerhub so, I'll talk to @jemc and @SeanTAllen when this is ready 😉 |
Ok, it seems the existing ubuntu base images are used with the various LLVM versions we're supporting. Would we want to multiply that -- i.e. replicate the LLVM dockerfiles that are built with Happy to help, though; but I think I'd rather get this merged and look into adding ubuntu 19.04 ci coverage separately. 🤔 |
ℹ️ Please don't merge -- I'm not done reproducing the original issue, and verifying that this solves it. (In the narrowest scope, at least, this will make |
😃 It's not resolved by this change alone, but I'm onto it. 🤞 |
I've taken a bit of a detour in 0de999d -- but since what all the
However, this seemed a bit disingenuous since that version part of the option really doesn't have much of an effect: It doesn't matter if you provide openssl_1.1.0 or openssl_1.1.1. So, I think I'd rather expose an option that reflects that better. What do you think? 😃 |
I've tried verifying that this now runs tests successfully on Ubuntu 19.04 using a docker container, but I keep running into #3099. Just spent some time trying with the vendored LLVM, but the same thing occurred. I might have used the wrong commands, anyways. My dockerfile:
The command I've run:
The output I see:
|
We've previously decided that we support Ubuntu LTS releases. That goes for CI so supporting Disco via would be a change that needs to be discussed. |
@SeanTAllen 👍 I'm fine with not pulling 19.04 into our CI. I'm just conflicted with this PR itself:
So, as a way forward, we could either merge this, to allow folks to provide |
@srenatus I think hardcoding in 1.1.1 for now and then we can figure out a longer term change makes sense. |
@srenatus yeah we discussed at sync and adding in hardcoded 1.1.1 as another option in Makefile is fine. Please let us know when this is ready for merging. |
I think this is good to go. I'll rebase, and see what CI thinks. |
…ions This introduces an array of valid versions, and only errors out if the passed default_ssl version is not among them. Unfortunately, the list of valid default_ssl version in the help output does not use that variable. Also switches the default_ssl debug output to "info" (was "error"). ----- In the stdlib code, this commit renames openssl_1.1.0 to openssl_1.1.x: Since the treatment is the same, we might as well call it what it is. Apparently it's confusing to set default_ssl to openssl_1.1.0 when you only have 1.1.1. So, let's simplify things by calling it openssl_1.1.x. ----- In Makefile-ponyc, we now treat openssl_1.1.{0,1} as openssl_1.1.x accordingly. Also removes the need for default_ssl_valid. Note that the introduction of pony_default_ssl was necessary; resetting default_ssl to some over value is _impossible_ when it was provided on the command line, e.g. `make default_ssl=something`. See https://www.gnu.org/software/make/manual/html_node/Overriding.html for details: > If you specify a value in this way, all ordinary assignments of the > same variable in the makefile are ignored; we say they have been > overridden by the command line argument. Signed-off-by: Stephan Renatus <[email protected]>
Thanks @srenatus |
[skip ci]
This introduces an array of valid versions, and only errors out if the
passed default_ssl version is not among them.
Unfortunately, the list of valid default_ssl version in the help output
does not use that variable.
Also switches the default_ssl debug output to "info" (was "error").
Fixes #3150, I hope.