[Top] [All Lists]

Re: WGLC for Sieve Notify has ended

2007-01-16 13:33:30

On Tue, Jan 16, 2007 at 12:19:00PM -0500, Mark E. Mallett wrote:

On Tue, Jan 16, 2007 at 11:27:54AM +0000, Alexey Melnikov wrote:

I will post the summary of all issues raised later this week.

I've only seen comments from Michael, Aaron and Dilyan. Has anybody else 
reviewed the document?

Yes... but too late, I fear.

Sorry for being abrupt.  I just have some probably-inconsequential

   Usage:  notify ":method" string
           [":from" string]
           [":importance" <"1" / "2" / "3">]
           [":options" string-list]
           [":message" string]

If I interpret the past discussion correctly, this may be getting
changed to remove ':method' and add a positional <method: string> .
I think that is a positive change, but I don't feel too strongly.

4.  Extract_text Action

   Usage:  extract_text [MODIFIER] [":first" number]
           <varname: string>

I feel that this would be much better as a standalone extension, or at
least as part of an extension where it is more relevant (e.g. "mime").
It has broader use than for just notify; users might want it when notify
is not available; and it could be specified in a more general way.  I
also think that notify could easily proceed without it (though I don't
know all the history, so maybe I'm wrong).  Having it here has the same
feel as a legislative earmark.  This is the strongest I feel of the
remarks that I have here :)

There's also this:

   The Extract_text action stores at most :first bytes of the first
   text/* part in the variable identified by varname.  If the :first
   parameter is not present, the whole content of the first text/* part
   is stored.

Should something be said about the upper bound?  If nothing else, that
the "whole content" may not be stored because of the truncation that
happens when the variables limit is exceeded.  I also think it would
help to say that implementation limits may apply.  My implementation,
for example, caches in memory some minimum number of bytes of each mime
part, with any excess possibly being cached on disk.  I'd happily make
what's cached in memory available, but I might be disgruntled by having
to fetch more from disk cache.  (This would probably already exceed the
variables limit, but it's just an example of why an implementation might
want to have a limit.)

5.  Test valid_notif_method

   Usage:  valid_notif_method <notification-uris: string-list>

This has the feel of "creat()".  i.e., are we so short on space that
we can't have the 'y' ?

Guess that's it..