Re: decode in mail

[Alain Magloire]

> 
> 
> Bonjour,
> 
> Returning to our discussion about message sets :^)

Nice work here, let me update and see the result.  I'll
more comment a little later during the week.

> 
> (or a message part), so no need for recursion occurs. For example, the
> following specification (which is, surely, absurdly complex ^:)
> 
>       1-2 [3-4, 5 [4 [(3,4) [4-8], 5, 7-11 ]]]

8-), Hopefully no one in there right mind will send such a beast.

> 
> 1. An arbitrary string in the message spec is taken to mean the match on
> sender's name. Still this search is not done, because as the comment in the
> original util_expand_msglist() stated:
>         /* FIXME: all messages from sender argv[i] */
>         /* Annoying we can use address_create() for that
>            but to compare against what? The email ?  */
> 
> Really, what should we compare it against?

when the msglist isalpha(), it means:
user                All messages from the given user

The problem is to compare to which field?
for example

Sergey Poznyakoff <address@hidden>

? from far  # I was looking for Farrah

Do we search the personnal name, address_personal () --> "Sergey Poznyakoff"
Do we search the personnal name, address_local_part () --> "gray"
etc ...

Or the entire email string:
address_to_string() --> "Sergey Poznyakoff <address@hidden>"

It was not clear for me.


> 
> 2. A /../ specification (match on the subject line) simply does an
> strstr comparison. What if we implement regexps?

It can be an interresting GNU extension.  Not to hard to implement, albeit
the platform provide POSIX regex API, if not default to mailutils/lib/regex.c
wich is GNU regex.

> 
> PS: How is it with libmu_scm? 

No go.


--
alain