Hi Tomasz, On 30.01.2012 10:21, Tomasz Bursztyka wrote: >>> My proposal is about the contrary: there will be dozens of application >>> using session api, so why not limiting the information flow from dbus to >>> the strictly necessary only. >> Of course we should not send a new message just to clear the settings, >> this can be as is it now send in the same message. > My point was not about multiple message but about message's content. But > ok I am fine with how it's done now, though I still think ui could be a > bit smarter. So let's forget patch 2. patch 6 is valid but need a small > modif (it will notify empty Ipv*)
I see. If we hit some problems with the current approach we can still change this later. This API is still experimental :) >> I think this is breaking the behaviour pattern of the ConnMan API we >> have so far. > Ok. I had in mind that, despite Manager/Service API, Session API will be > in use by dozens of apps so this will generate way more messages with > quite some content. My argument here is that if we start to optimize on this level (message size) we need some real numbers first. For example, how much do 50 extra bytes hurt? >> If it helps, I have a test UI for the sessions. With this UI you find >> easily bugs :) > Definitely! I am currently using test-session and some client of my own > which supports only a few parts of session API. > So your UI would really help here. I have a stand alone version of for the session api here[1] and I have started to integrate into my test connman ui [2]. If you want I can push the needed changes on the test ui for the api changes in this patch series. cheers, daniel [1] http://git.bmw-carit.de/?p=session-ui.git;a=summary [2] http://git.bmw-carit.de/?p=imposter.git;a=summary _______________________________________________ connman mailing list [email protected] http://lists.connman.net/listinfo/connman
