sieve lunch meeting notes

Our Sieve lunch meeting was surprisingly productive. Many drafts were 
discussed---pretty much the only thing we missed was "notify".
We danced around variables a lot and there is deep suspicion of variables 
but also a clear idea that they could make certain operations significantly 
easier.
Larry

New Conversation at: 3/20/2003 11:44:52 AM

[11:44]leg has arrived.

[11:47]kmurchison has arrived.

[11:47]<leg> New proposal for examining MIME headers. This is a separate from BODY

[11:49]<leg> Talking about Binary filtering (sorry, can't be more precise)

[11:49]<leg> Tim: this is not a good idea

[11:50]<leg> Searching for binary data encoded in base64 is ugly: need 3 combinations for different offset of NUL character

[11:51]<leg> BODY extention is good for last call, but need to add comparator so it can handle binary comparison

[11:52]<leg> Make comparator mandatory

[11:52]<leg> Ned: let's talk about spam test

[11:53]<leg> Do we want to add capability to return spam categories and list of rules as used by some spam tools?

[11:58]<leg> Modify spam test to return number (1 to 10) and text string. Text string is implementation defined and can contain spam category or something else

[11:58]<leg> Do we need variables in SIEVE.

[11:59]<leg> This would be great for spamtest

[11:59]<leg> Ned: Vacation draft

[12:00]<kmurchison> give time a "poke" for me :)

[12:00]<kmurchison> ^time^tim

[12:01]<leg> Ned: Change list of headers: remove BCC, add Resent-To, Resent-CC

[12:02]<leg> Tim: he is saying "hi"

[12:02]<leg> Jutta: want to send From address for vacations

[12:04]<leg> s/send/set

[12:04]<kmurchison> envelope or header?

[12:05]<leg> Larry was talking about CMU hack, but you probably know about it

[12:05]<kmurchison> yes

[12:05]<leg> Header in vacation reply

[12:05]<leg> Ned: let's talk about regex

[12:05]<kmurchison> il8n issues

[12:06]<leg> Ned: I have some code for posix regex and can share it

[12:06]<kmurchison> great. how do we handle il8n in the text?

[12:08]<leg> Nobody knows, but we have Ned

[12:08]<leg> 's support

[12:08]<kmurchison> also, people have expressed interest in backreferences (y/n)?

[12:09]<leg> Sorry, can you elaborate?

[12:10]<kmurchison> i think this would dovetail with variables. match part of an address and then use a backref variable as part of fileinto, or something

[12:10]<kmurchison> ask Simon, if he is there (looking at mailing list archives)

[12:11]<leg> Larry: regex without variables is bad

[12:11]<leg> Tim, Jutta: we disagree

[12:12]<leg> Tim: we can reserve $1 .. $9

[12:12]<leg> ... for regex

[12:12]<kmurchison> it seems to work for me without variables :)

[12:13]<kmurchison> shouldn't $1..$9 only be reserved under the scope of a regex? or is that too confusing?

[12:13]<leg> Dave Crocker: don't reinvent the wheel

[12:14]<kmurchison> can you elaborate on Daves's comment?

[12:14]<leg> There is no consensus on $1 .. $9 yet

[12:14]<leg> Dave is commenting about hot discussion about "variables: good or evil"

[12:15]<kmurchison> do we have consensus on strict POSIX regexes? ie, we don't allow \w, \s, etc?

[12:15]<leg> No idea

[12:16]<leg> Decision: regex is not ready, take to the list

[12:17]<leg> Ned will help you with i18n.

[12:17]<kmurchison> ok

[12:18]<kmurchison> what about subaddress? is it a keeper? is it ready for last call?

[12:19]<leg> subaddress is the next on the list. Just wait a second

[12:20]<leg> Ned: let's talk about subaddress. I implemented it.

[12:21]<leg> Ned doesn't remember any comments on subaddress

[12:21]<kmurchison> either do I

[12:21]<kmurchison> there was discussion on the usefullness of :detail

[12:21]<kmurchison> it migth be useful in the presence of variables however

[12:22]<leg> The only pushback can be on describing which character is delimiter ("-", "+", ...)

[12:23]<kmurchison> do we want it standardized, or configureable (either in the script, or byt site/implementation)?

[12:24]<leg> Yep, but nobody want's to write "variable" draft. Jutta has a draft

[12:25]<leg> Flags draft: remove mark/unmark

[12:25]<leg> Post your question to the list.

[12:25]<kmurchison> ok

[12:26]<leg> Problem: associating flag state with keep/fileinto

[12:28]<leg> addflags/removeflags is ugly in UI

[12:29]<leg> Proposal to drop addflags/.. also, preserve :globalflags only

[12:31]<leg> Coming back to "do we want it standardized ..." question. If your were talking about delimiter character, just tell that it is implementation dependent and people believe your text is good as is (?)

[12:32]<kmurchison> ok

[12:33]<leg> Variable draft will fix addflags/replaceflags

[12:34]<leg> Managesieve: ready for last call?

[12:35]<leg> Larry: Managesieve has defined a new URI scheme. I am confused though

[12:37]<leg> Why not use LDAP instead of Managesieve? Rant, rant, rant ...

[12:37]<kmurchison> i believe there are outstanding text fixes need w.r.t STARTTLS and auto-capability response (tmartin is aware of these)

[12:38]<kmurchison> i also believe that Rob and I extended the URL syntax

[12:39]<leg> Send changes to URL to the list, please.

[12:39]<kmurchison> ok

[12:43]<leg> Jutta: add parameter to fileinto: if the folder doesn't exist, create it.

[12:43]<leg> Ok

[12:44]<leg> Ned: Add Date extention: date in date/received header or current date

[12:44]<leg> Add comparator for date?

[12:45]<leg> Also need timezone information

[12:52]leg has arrived.

[12:53]<leg> Sorry, laptop powered off. But we are done anyway!

[12:56]kmurchison has arrived.