mysterious wordexp failure

[H.J. Lu]

> 
> 
> wordexp has been failing one of its tests for the past few days:
> 
>     Test words: <$(echo $var)>, need retval 0, wordc 2
>     Got retval 0, wordc 1: < abc >
> 
> where $var has previously been set to "abc".  Wordc 1 seems like the
> right answer to me, but neither the test nor the code has been changed
> since last October.  What gives?
> 

I used egcs in CVS at 08:45am on Apr. 2. I have no problems.

-- 
H.J. Lu (hjl@gnu.org)