Re: [Nmh-workers] problem with mark zeroing out sequences

[Ken Hornstein]

>>Of course, if you want to CALL mark inside of a hook, then all bets are
>>off :-)  I'm unclear how we can make that better.  I will note that rcvstore
>>can add messages to specific sequences, and there was a deprecated feature
>
>Nice! If that happens after add-hook we're 90% there.
>The only problem is in this use case rcvstore is being called by slocal,
>so how do you tell rcvstore what -sequence is?

Well, instead of specifying a folder name, you could pipe the message
into rcvstore using "|" in your maildelivery file, that would let you specify
the command completely and add a -seq argument to rcvstore.

>Actually, from a cursory review of folder_addmsg, it looks like moving
>
>        if (unseen) {
>            set_unseen (mp, msgnum);
>        }
>
>to after the hook invocation should clear things right up...
>thus nmh commands can act on the being-added message without altering
>its sequence status. I might have to make that tweak and see.

I can save you the trouble; that's not going to change anything.  All
set_unseen does is modify the sequence status bit vector in the folder
structure.  The locks don't get released until seq_save() is called.

--Ken