This is a regression versus 2.78, where I note that various "--disable-clamav" flags were passed along to the makefile if ClamAV was NOT selected in the config. In this version, NOT selecting ClamAV results in behavior where mimedefang will build to REQUIRE clamav but not use it as a DEPENDENCY which is ... pretty obnoxious at best. Recommend gleaning the working parts of the Makefile from 2.78.
Hello. Thanks for informing me, but I did not get it. Could you please provide some additional details i.e. logs, make config output etc. And what exactly you want me to remove from Makefile
It's something to add [back] to the Makefile, the missing options. Specifically, this clause was in the 2.78 Makefile: .if defined (MIMEDEFANG_DISABLE_CLAMAV) || ! ${PORT_OPTIONS:MCLAMAV} CONFIGURE_ARGS+= --disable-antivirus \ --disable-clamav \ --disable-clamd .else BUILD_DEPENDS+= clamscan:${PORTSDIR}/security/clamav RUN_DEPENDS+= clamscan:${PORTSDIR}/security/clamav .endif In the new Makefile, this has been replaced with the inexplicably hard-coded: CLAMAV_BUILD_DEPENDS+= clamscan:security/clamav CLAMAV_RUN_DEPENDS+= clamscan:security/clamav CLAMAV_CONFIGURE_ENABLE+= antivirus clamav clamd Though the config hasn't had the "ClamAV" option removed, so I assume it's an oversight, and not a campaign to force people to install and use ClamAV.
Hello, I have just tried to build the port without ClamAV and it is perfectly fine. At least in poudriere Could you please provide some logs with an actual error message?
Yes, it builds fine, but it creates a RUNTIME dependency on ClamAV... which, of course, means that Mimedefang doesn't work unless ClamAV is installed. Steps to reproduce: 1) Ensure ClamAV is NOT installed on your system 2) UNcheck the "Enable Clamav" in mimedefang make config 3) mimedefang make install 4) try to run mimedefang, it will work 5) have mimedefang process something <- here's where it will fail Since I have included the COMPLETE FIX for the issue, and the steps above will reproduce it, I imagine this should be sufficient, but let me know if you still have trouble figuring this out.
Thanks for the clarification. I have figured out that entire CLAMAV_* section is noop. Will submit a fix for this tomorrow.
Created attachment 191030 [details] patch Please try the patch attached.
Reporter feedback please!
Patch (maintainer approved) looks good to fix the reported issue. Explicit confirmation of QA (poudriere) passing with expected behaviour in both CLAMAV option enabled/disabled cases would be great. Port bugfix, MFH
Well it appears this one was fixed already see ports r495764
Assign to committer that resolved Re-open for MFH (quarterly is affected)
(In reply to Kubilay Kocak from comment #10) 2019Q3 is not affected. ports r495764 was committed in March.