Bug 214256

Summary: security/p5-GSSAPI: conflict with poudriere and openssl from port
Product: Ports & Packages Reporter: Sergei Vyshenski <svysh.fbsd>
Component: Individual Port(s)Assignee: freebsd-perl (Nobody) <perl>
Status: Closed Overcome By Events    
Severity: Affects Some People CC: jcfyecrayz, svysh.fbsd
Priority: --- Flags: bugzilla: maintainer-feedback? (perl)
Version: Latest   
Hardware: Any   
OS: Any   

Description Sergei Vyshenski 2016-11-06 00:30:26 UTC 
- Can be tested OK at 11.0-amd64 
with poudriere-3.1.14, perl-5.24.1, and openssl from base.

- But if I choose openssl from ports, that is add a line 
DEFAULT_VERSIONS+=      ssl=openssl
to file /usr/local/etc/poudriere.d/make.conf,
then poudriere fails with:
===>  p5-GSSAPI-0.28_1 You are using OpenSSL from ports and have selected
GSSAPI from base, please select another GSSAPI value.
*** Error code 1
Stop.
make: stopped in /usr/ports/security/p5-GSSAPI

This problem looks like a poor design of port security/p5-GSSAPI, because it does not allow testing from within poudriere with openssl from ports.

This also prevents testing poudriere for all ports, that depend on this one.
Comment 1 Mathieu Arnold freebsd_committer freebsd_triage 2016-11-06 09:31:23 UTC 
It is a known problem, I have a redesign of USES=gssapi coming up, introducing a DEFAULT_VERSIONS+=gssapi=<foo>.
Comment 2 John Hein 2017-01-06 14:28:04 UTC 
I can still reproduce the original problem on 10.x.  Could you explain the "events" that closed this bug?
Comment 3 Sergei Vyshenski 2017-01-06 14:53:46 UTC 
As Mathieu Arnold pointed, this issue is a known part of more complex problem, which he is working at.
Comment 4 Sergei Vyshenski 2017-01-06 15:15:23 UTC 
Also, it turned out that you can pass to poudriere any config option for any port. See thread of Dec 25, 2016 on freebsd-ports:

https://docs.freebsd.org/cgi/getmsg.cgi?fetch=40695+0+archive/2017/freebsd-ports/20170101.freebsd-ports

This provides much more control on testing within poudriere, than I wanted while reporting this PR.
Comment 5 John Hein 2017-01-06 17:24:07 UTC 
Okay.  Thanks for the explanation - it's helpful to have context for future people who hit the problem (instead of just seeing a close with no explanation of the resolution).