OSX: fix OpenSSL detection for 10.11. Fixes #690 #691
Conversation
|
The checks that have failed are to do with a configuration problem on Travis. I don't have access to an Apple environment. @deanberris can you confirm that this can be integrated? Also, does this fix need to be applied on the master branch? |
|
What spurred my need for a fix was https://build.getmonero.org/builders/kovri-all-osx-10.11/builds/4 I can't easily provide a new fixed build report because I can't easily update our submodule to point to my branch without causing other issues. I did ssh into this machine though and confirm manually. |
|
AFAICT this will also be needed in master, I'll confirm right now. |
|
Yes, it's needed for master too. |
| if (APPLE) | ||
| # If we're on OS X check for Homebrew's copy of OpenSSL instead of Apple's | ||
| if (NOT OpenSSL_DIR) | ||
| execute_process (COMMAND brew --prefix openssl |
There was a problem hiding this comment.
I'm not a fan of requiring homebrew just like this unconditionally on APPLE. Personally, when I build the library I provide the OpenSSL_DIR option myself.
I'd be open to accepting this if you make it so that we can detect whether the user has homebrew installed (something that could be checked at configuration time).
There was a problem hiding this comment.
@deanberris Agreed. I'll make the changes. If possible, maybe I should only apply this to OSX 10.11 since other versions appear unaffected?
There was a problem hiding this comment.
I think optional homebrew applies even to before OSX 10.11.
|
@deanberris @glynos I've pushed acb020b. Its a little biased with the Homebrew warning but it will stay out of the way. Confirmed functional/build successful on 10.11 and earlier machines. Should I reopen this PR to master? |
|
Also, yes, please do the same to master -- that would be most appreciated. |
- Cherry-pick's acb020b 0.13-release - References cpp-netlib#691
This is a confirmed fix for #690 and also confirmed to work on 10.9.5 (this should work on 10.10 as well).
The problem with this fix is the assumption that the user is using homebrew. If we shouldn't enforce this requisite, I'm open to ideas for an alternate fix.