Support OpenSSL 1.1.0 #2223
Support OpenSSL 1.1.0 #2223
Conversation
| m_iPubKeyBits = EVP_PKEY_bits(p); | ||
| m_szPubKeyType = (p->type == NID_undef) ? __tr("Unknown") : OBJ_nid2ln(p->type); | ||
| m_szPubKeyType = (type == NID_undef) ? __tr("Unknown") : OBJ_nid2ln(type); |
There was a problem hiding this comment.
Not sure about this one. There is no longer a way to get p->type - EVP_PKEY_base_id(p) replaces EVP_PKEY_type(p->type), however I suspect that the original code was wrong and this was the intent.
CyberShadow
force-pushed
the
pull-20170524-043815
branch
from
31e3b8e
to
e6088e4
Compare
|
Travis currently bundles openssl 1.0.1f, so compilation failed because new functions implemented in 1.1 are used in this PR, eg. EVP_MD_CTX_new. The easiest solution is probably to KVIrc depends on Qt that is linked to OpenSSL itself; this needs a bit of investigation, will it work if KVIrc is linked to OpenSSL 1.1 while Qt still links to a previous version? |
CyberShadow
force-pushed
the
pull-20170524-043815
branch
from
e6088e4
to
c58fddf
Compare
Ah crud, those were renamed in 1.1.0. Fixed.
I don't know. You may want to detect that in the CMake script. With this PR, it builds and works fine for me (on Arch Linux). |
|
see this: mumble-voip/mumble@dff1557 |
|
Ping As I said above, everything works fine on Arch Linux. Any additional changes for distributions in which Qt links to an older OpenSSL would need to be looked into by someone else. |
|
Ping. Is there any actionable feedback to be addressed here? Are the Qt concerns theoretical, or is anyone actually having trouble with this patch? We shouldn't attempt to fix what we can't even reproduce. |
|
I agree this one could me merged, as it:
|
|
Thanks! |
Fixes #2151.