[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: delete no longer working



>>>>> In article <ilug1emr0tn.fsf@xiphias.pdc.kth.se>,
>>>>> Simon Josefsson <jas@pdc.kth.se> scribbled:

  [snip]

  > Yup. For now we only set the \Deleted flag on the article.

  > Previous we also did CLOSE (which EXPUNGEs articles) when quiting
  > a mailbox, perhaps we should go back to this behaviour.

That would work for me for now. ;-)

  > Tech talk:

  [snip]

  > The problem with EXPUNGE is that you can't be sure of what you
  > expunge, if the \Deleted flag should have any user-visible
  > semantics at all you could have articles marked \Deleted but keep
  > them in the mailbox. Then EXPUNGE can't be used, hence we cannot
  > delete articles at all.

I've just scanned RFC 2060 for discussion of the \Deleted flag, and it
seems to me that it would be very dangerous to assign any user-visible
semantic to \Deleted other than 'will-go-away-when-expunged'.  Which
means, I suppose that Microsoft and Netscape will both come up with
clients that have a dumb semantic.

  > One suggestion I read somewhere was to unmark all flags, mark the
  > article you wish to delete and then do EXPUNGE. This is
  > single-client oriented and I don't like it.

Agreed.

  > What am I supposed to do?  Either we can let the user have the
  > option of deleting a given article (Yiha!), or we can have the
  > \Deleted flag mean anything at all to the user, IMO we cannot have
  > both.

I'm willing to settle for a third variant, in which \Deleted is a
session mark and only gets applied at session end.  You know, the
earlier nnimap semantic. ;-)

  > Netscape and Pine seem to have choosen not to let the user delete
  > a given article. I'm not sure I like this.

  > Perhaps I haven't groked what the \Deleted flag is supposed to be
  > used for, or something...

I _think_ you did.  It seems to have been designed to work in a way
similar to the delete option on early Unix mailers -- or, for that
matter, the way dired works -- flags are set to indicate that the
operation *should* be performed eventually as a result of a specific
'do-it' command.

marty

-- 
     marty fouts     fouts@null.net     http://www.best.com/~mjf

The danger already exists that mathematicians have made a covenant
with the devil to darken the spirit and confine man in the bonds of
Hell.  -- St. Augustine