[53376] trunk/base/src/port1.0/portutil.tcl

[jmr at macports.org]

Revision: 53376
          http://trac.macports.org/changeset/53376
Author:   jmr at macports.org
Date:     2009-07-04 04:32:45 -0700 (Sat, 04 Jul 2009)
Log Message:
-----------
remove strange check from canonicalize_variants that prevented some platform variants from being recorded in the registry, and update comment

Modified Paths:
--------------
    trunk/base/src/port1.0/portutil.tcl

Modified: trunk/base/src/port1.0/portutil.tcl
===================================================================
--- trunk/base/src/port1.0/portutil.tcl	2009-07-04 08:53:17 UTC (rev 53375)
+++ trunk/base/src/port1.0/portutil.tcl	2009-07-04 11:32:45 UTC (rev 53376)
@@ -1719,18 +1719,13 @@
     # was turned on or off, a particular instance of the port is uniquely
     # characterized by the set of variants that are *on*. Thus, record those
     # variants in a string in a standard order as +var1+var2 etc.
-    # We can skip the platform and architecture since those are always
-    # requested.  XXX: Is that really true? What if the user explicitly
-    # overrides the platform and architecture variants? Will the registry get
-    # bollixed? It would seem safer to me to just leave in all the variants that
-    # are on, but for now I'm just leaving the skipping code as it was in the
-    # previous version.
+    # XXX: this doesn't quite work because of default variants, see ticket #2377
 proc canonicalize_variants {variants} {
     array set vara $variants
     set result ""
     set vlist [lsort -ascii [array names vara]]
     foreach v $vlist {
-        if {$vara($v) == "+" && $v ne [option os.platform] && $v ne [option os.arch]} {
+        if {$vara($v) == "+"} {
             append result +$v
         }
     }
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.macosforge.org/pipermail/macports-changes/attachments/20090704/5d70c027/attachment.html>