These are chat archives for abranson/rockpool

19th
Jun 2016
Andrew Branson
@abranson
Jun 19 2016 20:44
I didn't know there was any trimming done previously. I've had some pretty long notifications on the watch.
ruff
@rufferson
Jun 19 2016 20:45
512 chars is quite long
Andrew Branson
@abranson
Jun 19 2016 20:45
yes indeed. were you trimming at that before?
i've had empty messages, and my chars are plain latin
some calendar pins seem to be getting lost still too. i had five events yesterday, and on the watch I only saw one of them, and there were two instances of it.
ruff
@rufferson
Jun 19 2016 20:53
me - yes, original rockworkd is not trimming it. These missing calendar pins - are they present in timeline folder? If yes - with which flag combination (third line in file)
Andrew Branson
@abranson
Jun 19 2016 20:55
i'll have a look at that next time i get some events
ruff
@rufferson
Jun 19 2016 20:55
also two instances - both are in the timeline folder or just one?
ruff
@rufferson
Jun 19 2016 21:01
and btw while making timeline reset i realized we don't need calendar migration stub, more appropriate way is to resync calendar
so i'm going to remove calendaritem and migration section replacing it with resync
Andrew Branson
@abranson
Jun 19 2016 22:09
yes i wondered that - the calendar pins are just a cache of the phone calendar anyway
the strange events were yesterday so i don't have them anymore. not much in the calendar for the next couple of days, but i'll check soon.
ruff
@rufferson
Jun 19 2016 22:11
they should still remain in the timeline for at least one day
or you mean yesterday they were for yesterday? %)
Andrew Branson
@abranson
Jun 19 2016 22:13
ok they're still there. 2 pins
it was a facebook event though
ruff
@rufferson
Jun 19 2016 22:13
facebook is a pin? not notification?
Andrew Branson
@abranson
Jun 19 2016 22:13
facebook does events
there's a facebook calendar
ruff
@rufferson
Jun 19 2016 22:14
and it's pushing events to mkcal?
Andrew Branson
@abranson
Jun 19 2016 22:14
yup
synced
the dataSource is the same, but the guid and id are different
ruff
@rufferson
Jun 19 2016 22:15
ok, so guid is not preserved. but then they should have been duped even before timeline
Andrew Branson
@abranson
Jun 19 2016 22:15
could have been a migration problem though - i switched to your beta in the last week so one of them could have been brought over from the old calendar entries, and also imported
ruff
@rufferson
Jun 19 2016 22:15
or... no, calendarevent does that weird content comparison
Andrew Branson
@abranson
Jun 19 2016 22:16
updateTime is different
ruff
@rufferson
Jun 19 2016 22:16
although even content is diffed when guis matches
Andrew Branson
@abranson
Jun 19 2016 22:16
so one of them was at least 5 days old
i'll call that a migration problem unless it happens again
i can't see any flags in the other ones though
ruff
@rufferson
Jun 19 2016 22:17
yes, that what i think, event was uodated, that generated new event actually instead of updating existing. Now, the question is - was old event removed from cal? If yes - i think that's what we miss, cal pins removal
Andrew Branson
@abranson
Jun 19 2016 22:18
cal only had one event
ah you think the cal pins aren't being removed at all?
ruff
@rufferson
Jun 19 2016 22:19
yes, we're not validating cal pins whether they still exist in the cal
Andrew Branson
@abranson
Jun 19 2016 22:19
you know, i think I saw one that I'd already deleted from the calendar
i assumed my caldav sync was broken
ruff
@rufferson
Jun 19 2016 22:19
they are validated actually for as long as daemon runs. but when it restarts - track is lost
and it restarts quite frequently tbh
need to review that tracking code again
Andrew Branson
@abranson
Jun 19 2016 22:21
the old code compared everything
this one didn't appear at all on the watch:
ruff
@rufferson
Jun 19 2016 22:21
actually when event is updated that should trigger refresh, refresh should detect removal and recall the pin
Andrew Branson
@abranson
Jun 19 2016 22:21
2016-06-13T21:41:05Z
2016-06-13T21:41:05Z
1001
{
"actions": [
{
"title": "Open",
"type": "open"
},
{
"title": "Snooze",
"type": "snooze"
}
],
"createTime": "2016-06-13T21:41:05Z",
"dataSource": "calendarEvent:ed429c16-f674-4220-95da-454f303f15e2",
"guid": "5bcdb0c1-5c58-471a-a583-a5896be75e88",
"id": "5bcdb0c1-5c58-471a-a583-a5896be75e88",
"layout": {
"backgroundColor": "vividcerulean",
"headings": [
"Calendar",
"Comments"
],
"locationName": "Ferney",
"paragraphs": [
"Home",
"buteo:caldav:uri:/remote.php/caldav/calendars/andrew/home/5bcdb0c1-5c58-471a-a583-a5896be75e88.ics;buteo:caldav:etag:\"46bce2f47213ea7a69163e0773e182b8\""
],
"title": "Sound check",
"type": "calendarPin"
},
"reminders": [
{
"layout": {
"locationName": "Ferney",
"tinyIcon": "system://images/NOTIFICATION_REMINDER",
"title": "Sound check",
"type": "genericReminder"
},
"time": "Sat Jun 18 15:00:00 2016"
}
],
"time": "2016-06-18T13:30:00Z",
"updateTime": "2016-06-13T21:41:05Z"
}
ruff
@rufferson
Jun 19 2016 22:23
1001 means pin was deleted
Andrew Branson
@abranson
Jun 19 2016 22:23
ah ok, it shouldn't have been deleted though
the event was still in my calendar
ruff
@rufferson
Jun 19 2016 22:24
1 - sendable (valid) 0 - sent, 0 - rejected, 1 - deleted
Andrew Branson
@abranson
Jun 19 2016 22:24
ah so it was deleted on the watch?
or not even sent?
ruff
@rufferson
Jun 19 2016 22:25
well, hard to say, any attempt to delete it would set the flag and recall them from watch
could be initiated by cal, or by maintenance (out of window)
since there's no action remove - cannot be removed from watch
manually that is
Andrew Branson
@abranson
Jun 19 2016 22:26
well unless it was an unintentional watch button press, it wasn't deleted.
ruff
@rufferson
Jun 19 2016 22:26
or that action is dynamic, it's always prepended
yes, looking at code - remove action is dynamic for pins, always prepended
Andrew Branson
@abranson
Jun 19 2016 22:28
one of those facebook entries was 1100 too
ruff
@rufferson
Jun 19 2016 22:28
1100 should be on watches
Andrew Branson
@abranson
Jun 19 2016 22:29
i see
ruff
@rufferson
Jun 19 2016 22:29
but that's what i found today - some pins are disappearing when pebble is resetting (eg when daemon restarts and pebble doesn't get ack/nack it resets by watchdog)
hence i added that reset timeline button
Andrew Branson
@abranson
Jun 19 2016 22:30
ok, i'll update and see how it goes
ruff
@rufferson
Jun 19 2016 22:30
to make full reprovisioning of the timeline from scratch
Andrew Branson
@abranson
Jun 19 2016 22:30
seems like there are lots of people trying it out though, so we'll get those bugs
ruff
@rufferson
Jun 19 2016 22:30
need to minimize daemon restart to prevent it
Andrew Branson
@abranson
Jun 19 2016 22:31
can't rely on that - but it doesn't happen to me now since I leaked the call handlers
ruff
@rufferson
Jun 19 2016 22:32
maybe to split rockpool in two pieces, watch connection thin layer and timline/platform/etc thick layer. so that connection-wise it will always be one-to-one
Andrew Branson
@abranson
Jun 19 2016 22:32
that would be much easier to share with ubuntu too
ah no sorry, didn't catch your meaning properly
what do you mean?
ruff
@rufferson
Jun 19 2016 22:33
well, two daemons and ipc between them kind of
Andrew Branson
@abranson
Jun 19 2016 22:34
why?
ruff
@rufferson
Jun 19 2016 22:35
so that any high logic collapse would not affect watchconnection, and if backend crashes it would simply nack the call. that should prevent pebble watchdog reset
but that would be too complicated architecture, so far i don't have any vision in my head how to implement it
just throwing an idea
Andrew Branson
@abranson
Jun 19 2016 22:36
right, got you
ruff
@rufferson
Jun 19 2016 22:38
ideally it should really be a plugin/service to bluez5 %)
Andrew Branson
@abranson
Jun 19 2016 22:38
hehe
well, i should know a lot more about that soon
new job starts tomorrow - at Jolla!
ruff
@rufferson
Jun 19 2016 22:40
i like le stuff, bought nordic chip recently with nrf/le support and cortex-m onboard
Andrew Branson
@abranson
Jun 19 2016 22:40
that has to be added - you're getting devices now that only support LE
like the new pebbles!
ruff
@rufferson
Jun 19 2016 22:42
oh really? that's great, hope they'll handle current turmoil and get better, i really love their ideas and sw/hw design.
and hope they wouldn't become KGB department :)
Andrew Branson
@abranson
Jun 19 2016 22:42
guess i'll find out! i got the feeling there's a lot more going on than they've announced
heh. i think he would have taken over the company if that was the aim
it's very good that he's started a downstream entity for it
not that i have any more suspicion about the russians than i do about the americans...
because we know for a fact that they're mining our data
ruff
@rufferson
Jun 19 2016 22:44
well, with americans it's more or less clear, they don't deny they're bombing your home :)
Andrew Branson
@abranson
Jun 19 2016 22:45
hehe. very true.
hope you haven't been personally affected by all that mess...
ruff
@rufferson
Jun 19 2016 22:46
unfortunately i was, I'm originally from Donetsk
Andrew Branson
@abranson
Jun 19 2016 22:47
oh no, I can't imagine what that's like
friend of mine was on that plane that got shot down though
do you still have family there?
ruff
@rufferson
Jun 19 2016 22:49
yes, they refuse to leave, just moved a bit aside from the front line, because our home was destroyed
Andrew Branson
@abranson
Jun 19 2016 22:51
wow, well I hope it gets stable soon. how do you think it'll end up?
ruff
@rufferson
Jun 19 2016 22:52
no idea, russia still refuses to return the border, till they do, it will be a mess
Andrew Branson
@abranson
Jun 19 2016 22:53
do you think it'll end up in russia?
ruff
@rufferson
Jun 19 2016 22:54
scarcely, they don't need it, they have plenty of such cakes around them, crimea was just too and too strategical to leave it floating around as others
Andrew Branson
@abranson
Jun 19 2016 22:56
well, best wishes to all your friends and family. i hope they all make it through ok.
ruff
@rufferson
Jun 19 2016 22:56
thanks, i hope too
btw i've started making voice endpoint in the wip. it will be for next release most probably
Andrew Branson
@abranson
Jun 19 2016 22:57
ah to connect to saera?
that will be fun
ruff
@rufferson
Jun 19 2016 22:57
unless i finish it till the time we stabilize timeline for 1.0 %)
Andrew Branson
@abranson
Jun 19 2016 22:57
:)
ruff
@rufferson
Jun 19 2016 22:58
yes, thinking to make endpoint serviceless, it will just emit signals to pebble/core/platform and unless someone picks up the call - will just nack it
Andrew Branson
@abranson
Jun 19 2016 22:59
yep good idea
default platform implementation just nacks
ruff
@rufferson
Jun 19 2016 23:00
if there will be any handler subscribed to voice signals - it should respond to session request and start receiving audio stream and sending back recognition
Andrew Branson
@abranson
Jun 19 2016 23:00
ah i didn't know you could tell if a signal was connected or not
that's cool
ruff
@rufferson
Jun 19 2016 23:01
endpoint cannot check, but pebble class can, it will check if any callback is registered. or maybe it will just forwrd dbus signal and fire up a timer. if someone replies - it resets the timer. and timer just nacks on timeout
Andrew Branson
@abranson
Jun 19 2016 23:03
ah right it's two way. cool.
better get to bed anyway. will let you know if i make any progress.
who knows, i might propose some sort of platform level smartwatch api!
ruff
@rufferson
Jun 19 2016 23:04
right, ok, need to sleep over the idea as well %)
good luck, you have experiecne in that already %)
so it woudln't be from the wilds