-
-
Notifications
You must be signed in to change notification settings - Fork 12.5k
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
salt 2019.2.2 #45895
salt 2019.2.2 #45895
Conversation
test step failed on Catalina build. 16:51:33 Error: salt: failed |
1fa258e
to
5dd053e
Compare
I have added a workaround for broken + # Workaround for https://github.com/saltstack/salt/issues/55084
+ # Remove when fixed
+ inreplace "salt/utils/rsax931.py",
+ "lib = find_library('crypto')",
+ "lib = '#{Formula["[email protected]"].opt_lib}/libcrypto.dylib'" What do you think guys @Bo98, @fxcoudert? |
I'd prefer if they could fix the issue in their own way, rather than us forcing a hard path. I've commented on the upstream issue. |
On the other hand, if this is blocking Python 3.8.0, then let's get it fixed, indeed |
It's a shortcoming of import ctypes
ctypes.find_library('crypto') The problem is all this is at runtime and not build time, so there's no flags or anything we can use to control it, unless upstream completely rewrites it to detect and insert the paths at build time. |
Oh wait, just saw you commented the same upstream. Sorry for repeating! |
High Sierra:
Will updating M2Cyrpto help, or is that problematic? |
Have tried it already with no lack. |
Hmm that's concerning. We seemed to manage to build a bottle a month ago. I don't have a High Sierra machine to narrow it down much. |
Does a passing the verbose flag |
Doesn't seem so: salt.log Looks like only python formula changed since the last successful build (3.7.4 -> 3.7.5). Will try to find it out later. |
@Bo98 curiouser and curiouser, it works with reverted Homebrew/brew#6749 |
Not sure what on earth swig is doing. Can you try modify it so that it is |
Nope, this didn't help |
Didn't think it would (given it works absolutely fine on Mojave and Catalina) - but I just waited to make sure. I suppose the Homebrew bug before was masking something going badly wrong here - and we now see it. Do you have a |
Yep, I have exactly the same thoughts. Here it is, for the latest brew:
And for the brew with reverted Homebrew/brew#6749:
|
The difference here will be that Can you tell me the output of |
Latest:
Without Homebrew/brew#6749:
|
I'm surprised swig even worked before. What it like outside |
|
Btw, why for |
Huh. Interesting ordering. Didn't expect
On Mojave it lists the 10.15 SDK. It's because Homebrew picks the latest SDK, and Xcode 10 contains the 10.14 SDK. It can work ok for certain framework stuff (basically whatever was in the SDK prior to the However I don't know if that's necessarily the issue here given it builds fine on Mojave, but you can try tweak it - do |
This didn't help, but |
5dd053e
to
06cd2af
Compare
That's interesting to know - I'll keep that in mind for swig. I wholly expected |
CI is happy now. Also, probably it is a good idea to check all formulas with
and all swig formulas then:
|
Fair point. |
Are we ok to merge the workaround @fxcoudert? In the hail of Python 3.8 |
Created with brew bump-formula-pr.
saltstack/salt#55084 (comment)