ports/135466: portupgrade leaves orphaned lock files since ruby 1.8.7p160 upgrade

[Virgil Champlin]

>Number:         135466
>Category:       ports
>Synopsis:       portupgrade leaves orphaned lock files since ruby 1.8.7p160 upgrade
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    freebsd-ports-bugs
>State:          open
>Quarter:        
>Keywords:       
>Date-Required:
>Class:          sw-bug
>Submitter-Id:   current-users
>Arrival-Date:   Thu Jun 11 07:00:05 UTC 2009
>Closed-Date:
>Last-Modified:
>Originator:     Virgil Champlin
>Release:        FreeBSD 7.2-RELEASE amd64
>Organization:
Stupidog.org
>Environment:
System: FreeBSD whisper.stupidog.org 7.2-RELEASE FreeBSD 7.2-RELEASE #0: Fri May 8 00:18:37 PDT 2009 root at whisper.stupidog.org:/usr/obj/usr/src/sys/STUPIDOG amd64

>Description:

	Since the ruby-1.8.7.160,1 upgrade, ports-mgmt/portupgrade
	utilities will fail to remove their lock files
	/var/run/pkgdb.db.lock and /var/run/portsdb.lock.  Subsequent
	invocations more than 2 minutes later will display the warning
	"** Stale lock file was found. Removed.". 

	This is caused by the failure of the new version of ruby to run
	the finalizers for the classes PkgDB
	(/usr/local/lib/ruby/site_ruby/1.8/pkgdb.rb - line 106) and
	PortsDB (/usr/local/lib/ruby/site_ruby/1.8/pkgdb.rb - line 156).
	Their sole function is to remove the lock files.

>How-To-Repeat:

	Any invocation of portupgrade with the current version of
	lang/ruby18 (ruby-1.8.7p160) should leave the aforementioned
	orphaned lock files. The following works well on my systems.

	     sudo portupgrade -f ports-mgmt/portupgrade

>Fix:
	
	I'm sorry but I know next to nothing about Ruby so this is well
	above me.  One obvious but marginal suggestion is to revert to
	the previous version of lang/ruby18 (ruby-1.8.7p72). This
	assumes it's a new bug in Ruby and needs a followup bug report
	submitted to them.  I would be happy to help but I'm too
	unfamiliar to make this call.  For all I know, it has something
	to do with with a subtlety of Ruby finalizers and it is now
	working correctly.

	Another suggestion is to review the locking strategy in the
	portupgrade suite and eliminate its dependence on finalizers.
	That is their sole purpose for these classes and coming from
	Java[TM], I'm inclined to avoid finalizers.  I would be glad to
	help but I wouldn't let me make the decision.

>Release-Note:
>Audit-Trail:
>Unformatted: