Fixed compile error for OpenSSL 1.0 systems (#3739)#3912
Fixed compile error for OpenSSL 1.0 systems (#3739)#3912aleks-f merged 1 commit intopocoproject:develfrom
Conversation
69d15c5 to
570ec33
Compare
|
Pull requests should go to develop branch |
570ec33 to
3ffdf14
Compare
|
Oops, my apologies. I've changed the Pull Request to base off of devel. |
|
Hello @aleks-f I don't think our patch is the reason for the checks failing. I think the checks in the devel branch were failing before we committed. |
3ffdf14 to
8cce9f5
Compare
|
Rebased with latest devel to see if that helps it pass checks |
8cce9f5 to
f51a034
Compare
|
There seems to be a bug with the github checks. |
f51a034 to
c296476
Compare
|
Hi! I think, that problem a little bit more complicated than this PR. Please look at #3939 |
|
Hello! Our patch does address the problem you mention. We use a C macro to test the OpenSSL version and handle both cases (one where the function returns void and one that returns an int). |
|
Hello @aleks-f Do you know if our patch is acceptable for the 1.12.5 release? |
Please, try build with openssl version 1.0.2. And you will see that in this version EVP_CIPHER_CTX returns void, and it will be a compile time error. |
|
Hello @bas524, We successfully compiled against OpenSSL 1.0.2k on CentOS 7. I agree that EVP_CIPHER_CTX_init returns void. I can't explain why you are getting compile errors on your Solaris system, but below I explain why our patch works: The first line Otherwise, we compile the version that returns void. This is true if OpenSSL version is 1.0.2, for example. |
c296476 to
51b79cb
Compare
51b79cb to
f055c50
Compare
I'm sorry. I was wrong. It was my inattention. |
Trying again -- not sure why it failed checks last time.