[14:49] hello people [14:50] hello [14:50] hello [14:51] hello [14:51] *** gmr_menta joined #kannel [14:53] *** igor_ivoilov joined #kannel [14:54] hi everybody [14:55] Hi all. [14:55] Aarno, can you grap up the word and discuss on the roadmap from your perstective. [14:55] ? [14:56] i think, even if this sounds partial, that we must build the 2.0 around the push [14:57] push ? [14:57] it is used by many application protocols [14:57] and wap gateway as such is very thin [14:57] only optimizations for wireless transport [14:57] I'll be a bit quite, since I have some chaos here right now, and don't want to abrupt you guys, so keep going... [14:58] how is push related to wap gateway? [14:59] i think we need to separate what is new for 2.0 and what must be improved. [14:59] additional service, but one needed by many application protocols, mms for instance [15:00] of course we must implement 2.0 wap gateway, too [15:00] who needs a 2.0 gateway ? [15:00] does anyone know how CMG's wap 2.0 gateway works ? [15:00] you said its an apache 2.0 proxy, no? [15:00] indeed [15:01] but wap1.0 and ppg is also apache instances [15:01] s/is/are [15:01] did they develop custom mod_ 's ? [15:01] yep, it seems so [15:01] their wap 2.0 seems an mod_proxy changed [15:02] but basis is wtcp and whttp ? [15:02] that makes a lot of sense :-) [15:02] besides some header removal, they just have the radius+auth addon, and gzip content negotiation activated [15:02] yep, if they call it wap 2.0 [15:03] so my idea for kannel 2.0 would be to also migrate it to apache mods [15:03] including wap 1.0 and mms code [15:03] not sure yet about how to do sms module [15:04] why move a perfectly good wap 1.x gateway into apache ? [15:04] because it's not an "perfectly good" wap gw ;) [15:04] and besides [15:05] we could benefit from a major rewrite and cleanup from the code [15:05] read: I need wap and mms code isolated to use in another place [15:05] code reuse is a separate issue IMHO [15:06] I see many phones stop using wap gateways, among them Nokia 3650,6220,sonyericssonp800,900 [15:06] what I mean is that if we start thinking about k2.0 architecture, we can think about splitting everything and make nice .so libraryes, and then implement it as apache modules [15:06] is there a future for wap protocols? [15:07] for sure, mainly for wap push i guess [15:07] I hope not, but what about MMS ? [15:07] there is future for push [15:07] wap gateway itself does just a wireless optimisation [15:08] does lots of other things as well, depending on the product [15:09] yes, legacy things [15:09] which are important because there are lots of old phones around [15:09] umm, things operators want - like radius authentication, ldap lookups, prepay checks, device capabilities etc etc [15:11] these are additional services, expect capbilities thing [15:11] of course very important ones, too [15:11] agreed, but they have found their way into wap 1.x gateways [15:11] and they are protocol independent, and will be required in 2.x gateways too [15:14] davi: MMS indeed uses WAP Push [15:15] 2.0 requires new protocol stack [15:16] or actuallu very old one: tcp and http [15:16] with wireless optimisation [15:16] application level things should be quite unaffected [15:16] right, and the optimisations are not generally available via the sockets api [15:17] its down to tuning kernel params, process params, win32 registry settings etc [15:19] yea, tcp level is a problem [15:19] http level less so [15:19] Theres very little with HTTP, only CONNECT and something else [15:20] paulkeogh: wap 2.0 ? [15:20] yep, OPTIONS and deflate [15:21] so, its not hard :-) [15:21] errr, why only CONNECT ? I have a apache+mod_proxy and a squid delivering "wap 2.0 gw" and I see regular proxy requests [15:22] speaking about specs only [15:22] of course you can add more services [15:22] no, i meant that the wap 2.0 WHTTP spec talks about connect specifically, and assumes support for everything else [15:23] davipt - does you wap 2.0 have any WTCP optimisations ? [15:24] gzip and X-* header removal only [15:24] I have to leave for some minutes, brb [15:25] and as soon as I activated gzip in my apache+mod_proxy, it became more alike to CMG's one ;) [15:25] brb [15:26] any estimation how much work wtcp reuires ? [15:27] i've never tried to code it, but i wrote a document that tried to specify how each wtcp optimisation is enabled on both win32 and rh linux [15:27] i can circulate it if you wish [15:28] yes, it would be very useful [15:28] ok [15:30] but i'm not sure wtcp is needed; wap 2.0 phones seem to work fine into "standard" tcp proxy stacks [15:30] then there are markup languages [15:30] all three of them [15:31] wml, xhtml and html right ? [15:32] wml1, wml2 and xhtml [15:32] ok, is there a WBXML encoding for xhtml in 2.0 ? [15:32] problem is that content translation may be required [15:33] is it not mandatory, but very useful [15:33] no, only wml1 is tokenized [15:35] but if people start implement wm2, wml2 <-> xhtml may be required [15:36] How much WML2 do you think they'll be ? [15:38] don't know, but wml2 is supposed to work with wml1 [15:38] i think if all wml2 devices also support xhtml, then wml2 will never happen ??? [15:39] yes, but requirements are either wm2 or wml1 + xhtml [15:40] but i agree that we can do without wml2 [15:41] we should do translation only if devices require it [15:44] then there is provisioning and user agent profiling. Do we need these ? [15:46] We only need to worry about UAProf if the gateway needs to transcode content [15:47] And UAProf is unworkable and RDF sucks. We should use WURFL. [15:50] I agree with you about rdf. WURFL is normal xml ? [15:50] yes, much easier to work with (and keep up to date) [15:53] and for push, and for mms, we need ota over http [15:54] no sms any more ;) [15:58] *** flashz joined #kannel [16:16] *** marciso quit Freenode : "Leaving"