[BUG]Celeste: Endless loop in #tokensIn: with special characters

Bernhard Pieber bernhard at pieber.com
Sun Sep 15 15:09:51 UTC 2002


Hi all,

I found a bug in MailAddressTokenizer. If you do the following with an
e-mail-address that contains a special character you get an endless
loop:

MailAddressTokenizer tokensIn: 'Bšrnhard Pieber <bernhard at pieber.com>'

(Alt/Cmd-. works as an emergency exit.)

This is called from Celeste when you send a message and you have a user
name which has special characters. (German Umlaut in my test.)

I am not sure what the correct fix is, i.e. which class should be
responsible for catching that, MailAddressTokenizer or Celeste? Celeste
does already ISO-encode the user name. I skimmed over RFC822 and read
RFC2047 and it seems to me that the first part of the user name should
also be MIME-encoded. That means that it has to be parsed first. Perhaps
the easiest solution would to split the user name into real name and
e-mail-address and let Celeste MIME-encode just the real name.

What do you think? Suggestions from IETF standards gurus are highly
welcome. ;-)

Bernhard Pieber




More information about the Squeak-dev mailing list