[10:50:21] timo.sirainen joins the room [10:52:34] cyrus joins the room [10:53:24] Aaron Stone joins the room [10:53:45] Barry Leiba joins the room [10:53:56] /? [10:53:58] /help [10:54:18] Aaron Stone has set the subject to: Sieve @ IETF 78 [10:54:34] cyrus leaves the room [10:54:38] cyrus joins the room [10:54:39] ah [10:54:45] Barry Leiba has set the subject to: Sieve meeting at IETF 78, Maastricht [10:56:29] Anyone listening on the audio? [10:57:00] yes [10:57:16] lots of background mumbling right now :) [11:00:33] Aaron Stone is now known as aaron.stone [11:00:33] aaron.stone is now known as Aaron Stone [11:00:56] Aaron Stone leaves the room [11:00:56] aaron.stone joins the room [11:01:10] .msg aaron.stone yo-ho [11:01:53] hmm stream stops sometimes [11:03:15] I'll be scribe [11:03:35] beginning the session at 4 minutes after 1pm [11:03:42] the note well is noted [11:04:08] we will spend 10 minutes discussing each item on the agenda. use it or lose it :) [11:04:20] agenda bashing: seeing none, we continue as planned. [11:04:21] aka joins the room [11:04:37] excitement, ManageSieve is now RFC 5804! [11:05:27] mike lvl is too high [11:05:27] alexey is waiting for the first errata to be published on managesieve, since it was sitting for so long, we're sure that nobody re-re-re-read it again. [11:05:58] alexey says he's seen 3 new managesieve implementations in the last year or two [11:06:09] now in RFC Editor Queue: Notary. [11:06:15] joseph.yee joins the room [11:06:26] WGLC: Include, and Notify-SIP. [11:06:41] Reviews, please. [11:07:07] alexey.melnikov joins the room [11:07:21] sb: is the mic level still too high? [11:07:35] if this is the mic; its ok [11:07:51] +same [11:08:07] i guess you were hearing alexey yell on the other mic [11:08:35] it was more like he was eating the mic :) [11:09:29] Cyrus relates a thread on the mailing list about include between barry and ned. [11:09:53] i've implemented include and multiscript together [11:09:56] Tony is at the mic. Says he and Ned might bring the multiscript draft back. [11:10:26] A few rumbles of implementations in the room of multiscript. [11:12:45] Cyrus discusses lack of implementation experience of include, in the presence of implementations of multiscript, and we may be missing out on important edge cases and use cases. [11:13:28] Cyrus points to Barry as shepherd to make a call on standard vs. experimental, etc. [11:15:01] Tony relates his implementation's script generator. It builds a single large script from script fragments. [11:15:55] Alexey says we discussed this at length at a previous IETF, there's lots of interest, push forward. [11:18:22] Cyrus recalls a conversation he and I had about interaction with ManageSieve. Question of whether updating an included script immediately becomes "activated", or if the active including script needs to be activated again in order to pick up the new version of the included script. [11:19:16] Alexey and Cyrus discuss details of this, requirements on clients to know about this, bytecode-compiled implementations, etc. [11:19:22] my implementation uses a compiled binary format and it is updated once one of the source scripts is changed [11:19:29] only the master script needs to be active [11:19:53] and no re-activation is necessary when an included script changes in the background [11:20:10] stephan: let's say you want to update 3 included scripts, and their interaction is not backwards-compatible with their previous versions. [11:20:30] during the course of uploading, you will be in inconsistent states. [11:20:36] how to handle that? [11:20:42] hmm [11:20:53] true [11:21:03] some atomicity needs to be inforced there [11:22:13] let's discuss this on the list, and see what we can figure out? [11:22:20] moving on: imap-sieve [11:22:24] yes, sounds good [11:22:27] barry at the mic [11:23:16] looking for reviewers, implementors of imap-sieve. [11:23:38] Takehito Akagiri joins the room [11:23:51] cyrus notes that the current draft expires tomorrow. let's get an updated draft, and last call. the document has been around for a while without new comments. [11:23:58] moving on: external lists [11:24:16] sent an e-mail on this one yesterday [11:24:37] about interaction of extlists with comparators [11:24:58] aka leaves the room [11:26:23] in the schema [11:27:34] barry says no comparators in external lists. [11:27:44] we throw the comparison over the wall to the sieve engine to figure out. [11:27:55] I agree [11:28:17] the tag: url references some server-side list and its search functions [11:28:39] mic too high still [11:28:51] (or too close) [11:29:46] sounds ok [11:30:27] much better :) [11:30:34] we're all remembering these discussions, and are pretty settled on tag: and the indication it gives that the magic exists behind the curtain on the server side. [11:31:50] moving on: regex [11:31:52] comments, plz. [11:31:56] moving on: notify-presence [11:32:12] Sebastian Hagedorn joins the room [11:33:14] this adds the ability to test if a user is present in the moment of delivery, e.g. to choose notification by sip or xmpp in real-time. [11:33:38] barry says, this doc seemed good to everyone when it was introduced, still looks good. ready for LC. [11:34:06] alexey says stagger the last calls [11:34:37] cyrus says nah, let's overload and just work through the docs as we can. (read: pull teeth to get reviews) [11:34:44] moving on: vacation-seconds [11:35:25] barry at the mic. it's pretty straightforward. we have vacation :days, we add vacation :seconds (which allows hours and minutes, obviously). looks good, so LC, too. [11:36:33] autoreply is an information document describing how to use presence, notify, vacation, etc. all together in an effective way. [11:36:57] if there are significant auto reply use cases not described, please let barry know! [11:37:02] moving on: convert [11:37:25] alexey at the mic, reluctantly, wondering which alter ego of his wrote this document. [11:39:05] this document describes a :converted argument to fileinto action, which will convert some mime-attachment from type a to type b. [11:39:22] needs security considerations, looping issues, etc. [11:40:08] separate command would be better yes [11:40:30] much like enclose [11:40:31] needs discussion on other actions it might apply to, e.g. redirect [11:41:08] cyrus recalls the replace action. [11:41:31] that's a better example [11:41:35] perhaps this would be a top-level action, convert, rather than an argument. [11:41:57] would make implementation easier (but it also influences tests like body) [11:43:05] alexey: i'm coauthor of several of these documents. that means st. peter going to get some work. [11:43:59] robins and alexey: and does a conversion failure cause the script to abort, or the original mime type to remain in place or...??? [11:45:00] Takehito Akagiri leaves the room [11:45:13] cyrus: what if there's an out-of-band definition of favorite conversions, so we use a tag: uri to pick those. [11:46:06] moving on: WG Futures [11:48:45] there are two needs: one is handled with external lists, giving access to the addressbook. the other is a set of tests to pull out particulars of a vcard, and for that we may need new tests. [11:50:45] cyrus proposes a generic xml test, allowing xpath queries [11:51:00] mihaly: general xml testing sounds great [11:51:46] cyrus point to alexey, says, write it! (or give ideas to someone to write up) [11:52:36] cyrus wonders if there are other xml payloads that we might use, making xpath stuff generally useful, should we go this path. [11:53:50] let's go to the list to discuss the scope further, and to find a potential co-author. [11:53:59] eh? [11:54:11] :) [11:54:19] alexey points at stephan.bosch, says you're writing lots of good implementations lately, wouldn't you like to draft this one... :) [11:54:43] not sure I have an application for it right now [11:54:52] moving on: document to describe how EAI/IDN interacts with sieve. [11:54:59] do we have issues here? [11:55:06] we think we will [11:55:15] we'll leave it on the future todo for now [11:55:19] moving on: benefits of sieve document [11:55:30] who's good at technical marketing? [11:55:57] *hears crickets* [11:56:20] those are not in the audio stream :) [11:56:44] To my opinion one thing is still missing in the Sieve world: a standard means for (virtual) users to get notified about runtime errors. Perhaps ManageSieve could be extended to provide access to the runtime log somehow. [11:56:45] cyrus has the goal of wrapping up a few more major documents, and once we do, if we're done, we skip these work items and close up shop. [11:56:56] Takehito Akagiri joins the room [11:58:01] stephan: i will channel that in a moment. [11:59:06] barry at the mic: what's the difference between draft standard and full standard? an implementation and interoperability report. what does this mean? well, we have some number of scripts that we expect to work everywhere. but are all scripts expected to work everywhere? we can prove this false easily. [12:00:21] barry: and so, in summary, i don't think it's worth it. [12:00:51] tony: this isn't a protocol, it's a language. interop of languages is different than protocols. [12:01:58] cyrus: i saw us writing a conformance test, a set of scripts and messages, and expecting that all implementations can run those. check for MUSTs and SHOULDs and make sure an implementation can get those right. [12:02:36] mihaly: brute force test by shoveling ietf archives into a sieve implementation :) [12:04:10] barry discusses the possibility that there are combinations of sieve actions and tests that can't actually work together, but we don't know it. [12:04:59] cyrus says if we focus on the base spec, it's a small enough set. [12:07:44] kurt: why not create a collection of messages and tests and define those as the test suite? i work on perl, and we have an input for each use case. [12:08:51] timo.sirainen leaves the room [12:10:19] discussion on the scope of an interop report. [12:10:48] my implementation uses a test suite. It's actually written in sieve itself with a large special testsuite extension. It supports testing saved and produced messages, checking results etc. [12:11:12] alexey will ask IESG what the scope of an interop report might look like. [12:11:20] barry volunteers to write up a test plan I.D. [12:13:31] Takehito Akagiri leaves the room [12:14:02] Takehito Akagiri joins the room [12:15:01] barry says: at IBM we email the script owner to tell them that there was an error, and their script is disabled until they go and fix it. [12:15:33] mihaly: i run a server that does a lot of conversions, we get a lot of errors, it would be terrible to overwhelm the user [12:16:10] barry is not volunteering for this idea: collect reports on how implementations handle this, and write up a best practices document. [12:16:47] hehe [12:16:54] ok, i will [12:17:14] FWIW, we run a cronjob that reports runtime errors to users by email, but "redirect" sometimes produces spurious errors that aren't the script's fault. [12:17:27] cyrus points to stephan and assigns him to ask around how errors are reported [12:17:32] well [12:18:50] mihaly asks if sieve could be used for instant message. cyrus says it'd be neat to use for calendaring. but, cyrus says, we're built around the semantics of email pretty heavily. [12:21:56] discussion on reputation tests [12:22:10] external lists [12:22:17] modify external lists from within the sieve script [12:22:38] spamtest will measure reputation and such out of band [12:22:44] s/will/could [12:22:52] any other business? [12:24:04] perhaps accessing external string data (so not lists) [12:24:09] clarifying next steps: we have two documents in LC. we will make more LC calls shortly. we need document author for XML-Sieve tests. we need an author for benefits of sieve, eai/idn. [12:25:03] who expects to be in Beijing? [12:25:12] we see seven hands [12:25:25] we will ask again on the mailing list. [12:25:34] and with that, we close this meeting. thanks everyone! [12:26:34] Barry Leiba leaves the room [12:26:43] Sebastian Hagedorn leaves the room [12:27:56] Thanks everyone. [12:27:58] note to self: references to "mihaly" are "michiel leenaars" [12:27:58] cyrus leaves the room [12:28:17] bye [12:33:39] stephan.bosch leaves the room [12:39:32] Takehito Akagiri leaves the room [12:56:12] aaron.stone leaves the room [12:58:38] alexey.melnikov leaves the room: Logged out [13:02:36] joseph.yee leaves the room [13:13:15] alexey.melnikov joins the room [13:13:54] aaron.stone joins the room [14:38:03] aaron.stone leaves the room [14:41:01] alexey.melnikov leaves the room [15:38:33] alexey.melnikov joins the room [15:56:30] alexey.melnikov leaves the room