Bugzilla Bug 131190

/usr/lib/catalyst2/modules/catalyst_support.py
588,589d587
<               elif len(cur_array) != 1:
<                       myspec[cur_array[0]] = cur_array[1:]
591,592c589,595
<                       print "\n\tWARNING: No value set for key:
"+cur_array[0]
<                       print "\tdeleting key: "+cur_array[0]+"\n"
---
>                       myspec[cur_array[0]] = cur_array[1:]
>       for x in myspec:
>               # Delete empty key pair
>               if not myspec[x]:
>                       print "\n\tWARNING: No value set for key: "+x
>                       print "\tdeleting key: "+x+"\n"
>                       del myspec[x]

I did not find any erro in these codes,but I really can not build my target by
these codes with no erro message. It's very strange that an exception occured
in del myspec[x] phase. I do not know why , but after having changed these
codes like above my catalyst can work well again. So I think it may be a bug
only in some special environment and unfortunately I got it.

Please use 'diff -au' to generate a patch.

--- catalyst_support.py 2006-04-25 14:53:06.000000000 +0800
+++ catalyst_support.py.change  2006-04-25 14:55:33.000000000 +0800
@@ -585,14 +585,11 @@

                if len(cur_array) == 2:
                        myspec[cur_array[0]] = cur_array[1]
+               elif len(cur_array) != 1:
+                       myspec[cur_array[0]] = cur_array[1:]
                else:
-                       myspec[cur_array[0]] = cur_array[1:]
-       for x in myspec:
-               # Delete empty key pair
-               if not myspec[x]:
-                       print "\n\tWARNING: No value set for key: "+x
-                       print "\tdeleting key: "+x+"\n"
-                       del myspec[x]
+                       print "\n\tWARNING: No value set for key:
"+cur_array[0]
+                       print "\tdeleting key: "+cur_array[0]+"\n"
        #print myspec
        return myspec

Let me guess. You were getting a traceback with "RuntimeError: dictionary
changed size during iteration". The only time this code should have caused a
problem is if you have a malformed spec file, for example:

livecd/foo:
livecd/bar: a b c

I'm attaching a patch that should fix this properly.

Created an attachment (id=85459) [edit]
slight change to parse_spec

The original version used 'for x in myspec.keys()' but I changed it to 'for x
in myspec', but it apparently used .keys() for a good reason. The former
returns a list of keys. The latter creates an interator and goes over the keys.
The iterator gets very unhappy when one of the keys it's currently dealing with
disappears.

yeah!Thank you for your comment which make me know what's wrong with these
codes, I have just learned python for a shot time and I did not know the
difference between "for x in myspec.keys()" and "for x in myspec". In fact
these days I devote to using catalyst to produce my own livecd , so I have
changed some codes of catalys and defined some new variables in spec file.
These actions cause some problems which do not appear in common sense.
^^

This patch has been added now.  Thanks, Andrew!

Fixed in 2.0_rc45