Created attachment 155992 [details] shar with port Please, find attached shar with port. This gem required to bring latest gitlab to FreeBSD. I don't mind to be or not to be maintainer, so feel free take maintainerships if you want to.
While this gem is named "timfel-krb5-auth", it still has "krb5-auth" in it's .gemspec, breaking all the automagic in USE_RUBYGEMS. Alexander, could you have a look at this problem?
(In reply to Michael Moll from comment #1) > While this gem is named "timfel-krb5-auth", it still has "krb5-auth" in it's > .gemspec, breaking all the automagic in USE_RUBYGEMS. Alexander, could you have > a look at this problem? What is needed to fix this problem? Should i rename the "krb5-auth" to "timfel-krb5-auth"? If yerenkow@ has no time to solve i will try it.
Doh, this missed me. According to their sourcecode https://github.com/timfel/krb5-auth/blob/master/Rakefile#L17 PKG_NAME='timfel-krb5-auth' https://github.com/timfel/krb5-auth/blob/master/rubygem-krb5-auth.spec#L4 %define gemname krb5-auth So, there is no right name from their perspective, thus we can chose either name. Please Torsten, rename it to PORTNAME= krb5-auth and see if this solves everything, it should do.
Created attachment 157196 [details] updated shar with new PKGNAME I've changed the PKGNAME as requested. But i did not understand the problem and could not say, if it solves it. Please have a look therefore.
After some tinkering I have now something that's compiling, running and loading into Ruby. Will put up the review shortly...
I spoke too soon, fails in poudriere... also It seem slike some there's a potential problem with ports vs. base kerberos. Did this gem ever work on FreeBSD beyond the point of pure compilation?
I created this port to make gitlab be able to work. I didn't use kerberos, so this port is simply telling gitlab that dependency is satisfied. If it's not working, and conflicting with base kerberos, seems problem is similar to OpenSSL problem (base/port), maybe someone could give a hint? So, I see currently two ways - create this port as is, to help porting gitlab with less hacks. Or dig and wait until kerberos will work. I'm for first one :)
I wouldn't create a defect port just do satisfy the gitlab port. Especially since krb is a optional dependency and gitlab should just work fine without it. Therefore i would create it without kerberos support until we find a way to get it work.
I'm taking this since I have the other gitlab ruby port prs. Lemme know if you want it back.
Take.
no problem, just as a hint, I had a semi-working version by depending on security/krb5-112 and using this patch: --- ext/extconf.rb.orig 2015-06-01 15:52:13 UTC +++ ext/extconf.rb @@ -4,8 +4,8 @@ RbConfig::MAKEFILE_CONFIG['CC'] = ENV['C extension_name = 'krb5_auth' dir_config(extension_name) have_library("c", "main") -raise "Cannot find krb5 library" unless have_library("krb5","krb5_init_context") -raise "Cannot find krb5 headers" unless have_header("krb5.h") -have_library("com_err","error_message") -have_header("com_err.h") +raise "Cannot find krb5 library" unless find_library("krb5","krb5_init_context") +raise "Cannot find krb5 headers" unless find_header("krb5.h") +find_library("com_err","error_message") +find_header("com_err.h") create_makefile('krb5_auth') however, the installed module did not load correctly, probably because the wrong kerberos library was used at some point.
Going to take care of this.
back to pool.
A commit references this bug: Author: swills Date: Wed Jan 4 01:22:41 UTC 2017 New revision: 430526 URL: https://svnweb.freebsd.org/changeset/ports/430526 Log: security/rubygem-timfel-krb5: create port timfel-krb5-auth - ruby wrapper for KRB5 WWW: https://github.com/timfel/krb5-auth PR: 199695 Submitted by: yerenkow@gmail.com Changes: head/security/Makefile head/security/rubygem-timfel-krb5/ head/security/rubygem-timfel-krb5/Makefile head/security/rubygem-timfel-krb5/distinfo head/security/rubygem-timfel-krb5/files/ head/security/rubygem-timfel-krb5/files/patch-ext_extconf.rb head/security/rubygem-timfel-krb5/pkg-descr
Committed, thanks and sorry for the long wait.