On Sun, Dec 19, 2010 at 04:39:35AM -0000, Mutt wrote: > #3480: IPv6 literal email-address fails > ---------------------------+------------------------------------------------ > Reporter: Harry | Owner: mutt-dev > Type: defect | Status: new > Priority: minor | Milestone: > Component: mutt | Version: HEAD > Keywords: literal, IPv6 | > ---------------------------+------------------------------------------------ > Sending to a literal IPv6 address (test@[2001:470:ffff:aaaa::1], as an > example) will fail. > > How to reproduce: > 1) Make a new email To: anything@[2001:470:ffff:aaaa::1]", as an example. > 2) It gets transformed into the editor as "To: anything@[2001: 470: ffff: > aaaa: ;" - Even if the user corrects the malformed literal IP in their > editor, mutt will say "No recipients are specified!" when trying to send. > > Remco Rijnders informed me that it should be > "anything@[IPv6:2001:470:ffff:aaaa::1]", but even then it has the same > error (it malforms when mutt puts it into any editor). > > It would be nice if someone could look into this bug if they have time. > Thanks. Having had a further look at this, it seems the address parser / tokenizer sees ':' as a token to split on. I wonder if this is a widely used convention or one solely used by mutt. If there is no valid use case for seperating addresses with a colon, the literal IPv6 address problem can be easily fixed by taking the colon out of the tokenizer. If hower there is a valid reason to see the colon as a token seperator, more extensive logic and parsing will have to be applied to distinguish between colons used as token seperator and those colons used inside an IPv6 string literal address. Any thoughts or insights?
Attachment:
signature.asc
Description: Digital signature