Bug 243383 - devel/qca: Fails to build with libressl: error: static declaration of 'RSA_meth_set_sign' follows non-static declaration
Summary: devel/qca: Fails to build with libressl: error: static declaration of 'RSA_me...
Status: Open
Alias: None
Product: Ports & Packages
Classification: Unclassified
Component: Individual Port(s) (show other bugs)
Version: Latest
Hardware: Any Any
: --- Affects Some People
Assignee: kde
URL:
Keywords: needs-qa
Depends on:
Blocks:
 
Reported: 2020-01-15 22:46 UTC by dave
Modified: 2020-01-16 01:30 UTC (History)
2 users (show)

See Also:
bugzilla: maintainer-feedback? (kde)
koobs: maintainer-feedback? (brnrd)
koobs: merge-quarterly?


Attachments
Poudriere log of build (126.58 KB, text/plain)
2020-01-15 22:46 UTC, dave
no flags Details

Note You need to log in before you can comment on or make changes to this bug.
Description dave 2020-01-15 22:46:14 UTC
Created attachment 210780 [details]
Poudriere log of build

Building devel/qca with latest quarterly snapshot 2020Q1 on FreeBSD 11.3 with ssl=libressl and option SSL fails. The relevant log bits are at the end of the comment.

Note, this resembles but is not exactly equivalent to the following bugs:

bug #235883 
bug #232784

A complete log will be attached. 

In file included from /wrkdirs/usr/ports/devel/qca/work/qca-2.2.1/plugins/qca-ossl/qca-ossl.cpp:44:
/wrkdirs/usr/ports/devel/qca/work/qca-2.2.1/plugins/qca-ossl/ossl110-compat.h:260:12: error: static declaration of 'RSA_meth_set_sign' follows non-static declaration
static int RSA_meth_set_sign(RSA_METHOD *meth, int (*sign) (int type, const unsigned char *m,
           ^
/usr/local/include/openssl/rsa.h:480:5: note: previous declaration is here
int RSA_meth_set_sign(RSA_METHOD *rsa, int (*sign)(int type,
    ^
In file included from /wrkdirs/usr/ports/devel/qca/work/qca-2.2.1/plugins/qca-ossl/qca-ossl.cpp:44:
/wrkdirs/usr/ports/devel/qca/work/qca-2.2.1/plugins/qca-ossl/ossl110-compat.h:268:12: error: static declaration of 'RSA_meth_set_verify' follows non-static declaration
static int RSA_meth_set_verify(RSA_METHOD *meth, int (*verify) (int dtype, const unsigned char *m,
           ^
/usr/local/include/openssl/rsa.h:486:5: note: previous declaration is here
int RSA_meth_set_verify(RSA_METHOD *rsa, int (*verify)(int dtype,
    ^
/wrkdirs/usr/ports/devel/qca/work/qca-2.2.1/plugins/qca-ossl/qca-ossl.cpp:2849:4: error: call to 'RSA_meth_set_sign' is ambiguous
                        RSA_meth_set_sign(ops, NULL);
                        ^~~~~~~~~~~~~~~~~
/usr/local/include/openssl/rsa.h:480:5: note: candidate function
int RSA_meth_set_sign(RSA_METHOD *rsa, int (*sign)(int type,
    ^
/wrkdirs/usr/ports/devel/qca/work/qca-2.2.1/plugins/qca-ossl/ossl110-compat.h:260:12: note: candidate function
static int RSA_meth_set_sign(RSA_METHOD *meth, int (*sign) (int type, const unsigned char *m,
           ^
/wrkdirs/usr/ports/devel/qca/work/qca-2.2.1/plugins/qca-ossl/qca-ossl.cpp:2853:4: error: call to 'RSA_meth_set_verify' is ambiguous
                        RSA_meth_set_verify(ops, NULL); //pkcs11_rsa_verify
                        ^~~~~~~~~~~~~~~~~~~
/usr/local/include/openssl/rsa.h:486:5: note: candidate function
int RSA_meth_set_verify(RSA_METHOD *rsa, int (*verify)(int dtype,
    ^
/wrkdirs/usr/ports/devel/qca/work/qca-2.2.1/plugins/qca-ossl/ossl110-compat.h:268:12: note: candidate function
static int RSA_meth_set_verify(RSA_METHOD *meth, int (*verify) (int dtype, const unsigned char *m,
           ^
4 errors generated.
Comment 1 Kubilay Kocak freebsd_committer freebsd_triage 2020-01-16 00:12:42 UTC
Thank you for the report Dave

The log notes the system is running a 11.3-PRERELEASE, which is not (no longer) a  supported version for ports.

Can you test a poudriere build using an up-to-date 11.3-RELEASE-pX jail and report back as to whether the issue is reproducible there too please
Comment 2 dave 2020-01-16 01:30:54 UTC
First, while uname -says- it's PRERELEASE, please look at the revision number. I typically track -STABLE and have learned to ignore said designation ... preferring the svn revision number for more exactness.  

Next, according to what I know about poudriere, it will not build packages on OS jails that are higher revision than the current revision. Even if it can, somehow...if you are focused on the OS difference in the logs then you would logically also be focused on the OS difference between poudriere's jail and my own OS. :)

If there is a lower revision you would accept, please let me know.