Bugzilla – Bug 4833
SPEC -transition elements and/or next attribute
Last modified: 2013-08-28 22:08:24 UTC
Sorry for not directly referencing earlier bugs.. I know this was recently raised (the idea of having both <next> and @next attribute, and I shot down the idea.
However, since we had to revisit this in the RI anyway, I have a new opinion.
Here's what the spec should say, IMO:
1) First, we match against any of the transition elements.
2) Next, if we haven't matched, we follow the next attribute value, if one is specified
3) If no @next is specified, the job ends normally (COMPLETED).
Obviously it would be pointless to have the attribute take precedence over the transition elements...and I can't see a reason to exclude next as a transition element.
If there are no objections, can we please add this to the spec Chris?
Just to point out, SB considers the use of both the next attribute and the next transition element in the same step, etc to be semantically invalid. I'd have to double check the code but it may be an all or nothing thing (either you use the next attribute or transition elements in general).
This discussion has been superseded by discussion between Scott and Michael on the public ML with subject "evaluation order of multiple transition elements". Expect one or more new issues from that discussion.