Reported for v1.6.1.20 as it still occurs, but it happened as far back as v1.5.12.
With the following command name, VoiceAttack "freezes" while taking up more than a core's worth of CPU time and an increasing amount of memory. It's possible that when left running it'd eventually finish or outright crash, but the fact that it freezes is enough for me:
long dynamic number [1..2147483647]
You may note that this is the exact upper limit of an Int32. This is important because the following does not produce the same result:
long dynamic number [1..21474836470]
While the number range is much larger, it doesn't freeze, but also doesn't produce any command variations to the speech engine(as far as I can tell by speaking "long dynamic number one" and such). I would speculate it fails silently instead of attempting to actually process the range.
What does cause a freeze is
long dynamic number [1..214748369,10]
If I understand correctly, the final variant would be "long dynamic number 2147483690", which falls out of the maximum range an Int32, so it should either fail or crash at the end.
Finally,
long dynamic number [1..10,214748369]
doesn't freeze, but doesn't appear to generate anything either(Unrecognized : 'long dynamic number 214,740,369').
At first I figured something went wrong with the normal "Dynamic Command Generation Warning" message about too many variants, but the message does appear with a smaller range like "[1..2147]", so it's the chicken and egg problem(egg came first, proto-chicken laid it, but that's beside the point) of needing to process the variants before knowing how many there will be.
I'd also like to note that there is no option to heed the warning and cancel the creation of the command, it can only be acknowledged by clicking "OK", after which the command will still be generated.
And yes,
It's an edge case, I know