Re: [bug-mailutils] mail(x) message status bugs (with patch)

[Sergey Poznyakoff]

Hi Paul,

> 1.  The "preserve" command should do more than just undo the "mbox" command.
>     In fact, it needs its own status bit, since there are three possible
>     states associated with the preserve and mbox commands:  preserved,
>     mboxed, or neither (in the latter case, whether a message in the inbox
>     is moved to the mbox depends on whether the message was read in the
>     present invocation of mail).

Yes, this is reasonable. 

> 2.  Assuming there are no preserve, mbox, save, etc. commands, a message
>     in the inbox should only be moved to the mbox if it was read during
>     *this* invocation of mail.

I did not find any such requirement in POSIX.  It says that

   "As messages are read, they shall be marked to be moved to a
   secondary file for storage, unless specific action is taken."

but there is no mentioning whether this applies to the messages read in
this session or earlier.  Although from the logical point of view this
seems sound.

> 3.  The POSIX description of the "touch" command makes no mention of moving
>     a message to the "read" state, so the touch command needs its own
>     status bit.

OK.

> 4.  In all other versions of mailx that I've used, old messages that had been
>     read already are indicated in from/header lines with ' ' as the status,
>     not 'R'.  I find this more useful, since there are tons of such messages
>     in my inbox and this allows all other messages to stand out more.

Well, yes, but given that all previous versions of MU mail were using
this mark, we cannot simply get rid of it.  There may be some people
around there who's got used to it (as, e.g. yours faithful is:]), or who are
relying on it in their scripts (actually, that was the main reason for
introducing this mark).  I will make the exact appearance of the read
mark configurable via a special mail variable.

> 5.  The "undelete" command should unset a message's "preserved" and "mboxed"
>     status bits, since it is supposed to leave a message in the state "read".

OK

> 6.  If the current mailbox is not the system mailbox, then the "quit" command
>     should not delete saved messages.

Yes, certainly.

> I have attached a patch, and hope that you will consider including it in
> the next release of the Gnu mailtools.

Thanks a lot, I surely will.  But you meant GNU mailutils, of course :)

Regards,
Sergey