I did not concentrate enough on Frank Shearar message.
He got it right.

Color>>formString: implementationrely on the fact that an error is generated.

'16rYELLOW' asNumber did generate a bug in 3.8, no more in 3.9.

Note that this is half lucky, because in many cases, String>>asNumber is very permissive.
  '' asNumber = 0
  '16r' asNumber = 0
  '16rX' asNumber ERROR

This kind of weirdness motivated the design of SqueakNumberParser in 3.9 (hey, do not blame me, error pre-existed as a change in Integer>>readFrom:base: ).

SqueakNumberParser is able to raise an error, but default behavior is rather to answer 0 like previous 3.9 implementation...

Shouldn't this change in next version 3.10 or 4.0 ?


________________________________________________________________________
iFRANCE, exprimez-vous !
http://web.ifrance.com