**** BEGIN LOGGING AT Tue Aug 04 16:58:21 2020
Aug 04 17:01:14 <victhor> hi, what would be a good strategy for adding a new device to openwrt? I naively assumed I should base my changes on a stable version, and I wanted to submit my changes upstream, only to realize the mess I have on my hands, now that I have to rebase it to master... I figured starting out from stable made sense, as to avoid bugs that aren't hardware-related. should I branch off from master in the future, or maybe from the commit stable was branched
Aug 04 17:01:14 <victhor> off of?
**** ENDING LOGGING AT Tue Aug 04 17:02:41 2020

**** BEGIN LOGGING AT Tue Aug 04 17:16:18 2020
Aug 04 17:16:51 <jow> .../subscribe?path=foo;path=bar;typed=full
Aug 04 17:17:09 <jow> typed could be none (default), short (type only), full (object.type)
Aug 04 17:17:15 <rmilecki> so the basic format would omit object path
Aug 04 17:17:31 <rmilecki> as that's probably the most common use, right?
Aug 04 17:19:10 <jow> yeah, default should likely be "short"
Aug 04 17:19:50 <rmilecki> ok, just for sure: [default] event: status \n data: { "foo": "bar" }        (no object path available)
Aug 04 17:20:09 <rmilecki> did I get it right/
Aug 04 17:21:07 <rmilecki> in above "status" if event type and "data" is msg - as in
Aug 04 17:21:08 <rmilecki> int ubus_notify(struct ubus_context *ctx, struct ubus_object *obj, const char *type, struct blob_attr *msg, int timeout);
Aug 04 17:22:50 <jow> yes
Aug 04 17:23:03 <rmilecki> ok, perfect
Aug 04 17:23:24 <rmilecki> in that case, i'd call it another feature - support for GET /ubus/subscribe?....
Aug 04 17:23:32 <rmilecki> feature that probably can be added slightly later
Aug 04 17:23:43 <jow> and then two further choices: no event name at all (so client can use onmessage) or long event name (objectpath.type)
Aug 04 17:23:46 <rmilecki> in that case - is there anything else to improve / fix in my V2?
Aug 04 17:24:29 <jow> I think otherwise your V2 looked quite perfect already. Have yet to runtest it
Aug 04 17:24:37 <jow> just had another idea regarding the object path
Aug 04 17:24:46 <rmilecki> we can't use    event: objectpath.type   as that would make   event: hostapd.wlan0.status    unparse-able
Aug 04 17:25:16 <jow> we could send  id: objectpath@timestampmilliseconds
Aug 04 17:25:39 <rmilecki> is id available to JS?
Aug 04 17:26:35 <aparcar> jow: is there an elegant solution to build only 1 image instead of all? For the CI I don't want to create all ath79 images
Aug 04 17:26:39 <jow> yes, as eventSource.lastEventId
Aug 04 17:26:57 <jow> user code could use that to figure out the object path (after stripping the timestamp suffix)
Aug 04 17:27:19 <jow> instead of a timestamp you could also use a simple monotonically increasing counter
Aug 04 17:28:35 <rmilecki> jow: i think I like that!
Aug 04 17:29:22 <rmilecki> jow: should we make id objectpath@timestampmilliseconds OR timestampmilliseconds@objectpath ?
Aug 04 17:29:26 <jow> but not sure if lastEventId always refers to the event the handler function currently is running for or if it refers to the latest event received so far but before event handlers are fired, this is something that needs testing
Aug 04 17:29:34 <rmilecki> if you consider @ character meaning, later could make more sense
Aug 04 17:29:45 <rmilecki> if you think about hierarchy, first sounds better
Aug 04 17:29:55 <jow> not sure if we can use ms timestamps after all
Aug 04 17:30:20 <rmilecki> then objectpath@counter OR counter@objectpath ?
Aug 04 17:30:33 <jow> counter@objectpath
Aug 04 17:30:37 <rmilecki> ok
Aug 04 17:30:56 <rmilecki> jow: i really have to run now, should be home at least half an hour ago
Aug 04 17:30:59 <rmilecki> thanks for chatting!
**** BEGIN LOGGING AT Tue Aug 04 18:09:00 2020
Aug 04 19:13:19 <aparcar> adrianschmutzler: wb
Aug 04 19:14:14 <adrianschmutzler> will be gone again soon, just had to discuss something in another channel
Aug 04 19:19:43 <Borromini> :P
Aug 04 19:28:22 <aparcar> does the build system require GNU sed?
Aug 04 19:28:32 <aparcar> or any type of sed
Aug 04 19:31:07 <Borromini> https://openwrt.org/docs/guide-developer/build-system/install-buildsystem?#prerequisites
**** BEGIN LOGGING AT Tue Aug 04 19:56:21 2020
**** BEGIN LOGGING AT Tue Aug 04 21:11:47 2020
Aug 04 22:05:45 <aparcar> lynxis: ping
**** BEGIN LOGGING AT Tue Aug 04 23:53:34 2020
Aug 04 23:54:19 <russell--> can someone explain the differences between gpio-export and gpio-hog?
Aug 05 00:01:06 <russell--> ah, figured it out
Aug 05 00:01:25 <russell--> gpio-hog is for nailing a value, export if you need to change it
**** BEGIN LOGGING AT Wed Aug 05 02:49:22 2020
**** ENDING LOGGING AT Wed Aug 05 02:59:57 2020