| locusf | hi | 13:21 |
|---|---|---|
| tbr | lo | 13:21 |
| tbr | I've found a way to service MQTT PING messages properly in python | 13:22 |
| tbr | https://gist.github.com/dm8tbr/b36c6afca8904d15027a | 13:23 |
| locusf | ok | 13:25 |
| locusf | havent tried paho yet | 13:25 |
| tbr | it's mostly compatible | 13:25 |
| tbr | with the old (deprecated) python bindings | 13:25 |
| tbr | I plugged in an IPHB wakeup every 10s to make sure keepalives are serviced with at most 10s delay | 13:26 |
| tbr | and exactly that seems to happen and the server is happy | 13:26 |
| tbr | for style points I should probably trim the keepalive timer by those 10s | 13:27 |
| tbr | or maybe stretch it to 30s wakeup intervals and extend the general keepalive interval to something longer | 13:30 |
| tbr | ls | 14:17 |
| tbr | ok, I've pushed out a cleaned up version | 14:17 |
| tbr | beware, you'll have to edit mqtt_topic_base to match the right user name | 14:18 |
| tbr | I also need to package up paho... | 14:20 |
| tbr | with the iphb patch | 14:21 |
| tbr | which works lovely for me, I now get instant notifications, even if the device "sleeps" | 14:23 |
| kimmoli | oo | 14:36 |
| tbr | I also found the ubuntu-touch repositories, will need to look at what they did and if I can learn something from that | 15:04 |
| tbr | they decided against MQTT for weird reasons | 15:05 |
| kimmoli | i did read g+ | 15:42 |
| locusf | whats on g+? | 15:45 |
| locusf | oh tbrs post | 15:47 |
| tbr | yeah | 15:48 |
| tbr | I hoped to get something going, but they don't use their mailing list, at all. | 15:48 |
| locusf | tbr: is paho packaged or should I just install from sources? | 16:01 |
| locusf | tbr: already dl'd sources and python installed | 16:03 |
| tbr | locusf: I currently have it from sources | 16:03 |
| tbr | packaging is on my list | 16:03 |
| tbr | I should upload my 'fork' though | 16:04 |
| tbr | as it has the iphb patches | 16:04 |
| locusf | tbr: ok | 16:06 |
| tbr | git remote remove origin; git remote add origin https://github.com/dm8tbr/org.eclipse.paho.mqtt.python.git | 16:10 |
| tbr | and maybe some other incantation | 16:10 |
| locusf | iphb needs to be installed? | 16:11 |
| tbr | I'd be very surprised if it isn't :) | 16:11 |
| tbr | as it's part of the platform | 16:11 |
| tbr | I needed to set: git branch --set-upstream-to=origin/master master | 16:14 |
| locusf | hmm I don't receive even test notifications now even though I have changed topic | 16:15 |
| locusf | any changes needed to the perl side? | 16:15 |
| tbr | might be something gone wrong :/ | 16:20 |
| locusf | mqtt_notify_topic = sailfish/locusf/irssi/notifications | 16:21 |
| tbr | ah, no | 16:22 |
| tbr | sailfish/locusf/ | 16:22 |
| locusf | oh really? | 16:22 |
| tbr | wait | 16:22 |
| tbr | on the irssi side you're correct | 16:23 |
| tbr | and that should not change | 16:23 |
| locusf | /set mqtt_notify_topic sailfish/locusf/irssi ? | 16:23 |
| tbr | if you're not getting anything, then I screwed up the jolla side code | 16:23 |
| tbr | irssi is fine | 16:23 |
| tbr | I won't change anything there | 16:23 |
| tbr | have you set "mqtt_topic_base" to sailfish/locusf/? | 16:24 |
| tbr | in the python script | 16:24 |
| locusf | nope, its already set as "sailfish/"+mqtt_user | 16:24 |
| tbr | yeah, that won't work for you | 16:25 |
| tbr | as your user is locusf_read | 16:25 |
| locusf | oh ok | 16:25 |
| tbr | sorry, the user/client structure is a bit in flux as I'm trying different things to figure out a good approach for the platform notifications | 16:26 |
| tbr | my device currently has an account name == imei and a client id sailfish_$IMEI | 16:27 |
| tbr | as I'm considering the IMEI as the unique device identifier | 16:27 |
| locusf | hmm my irssi froze for a long time just now for just testing | 16:29 |
| tbr | that happens if mosqitto_pub stalls :/ | 16:29 |
| tbr | can you please set your timeout to 210s? your client goes amock and reconnects all the time | 16:30 |
| tbr | ah no timeout is right | 16:30 |
| tbr | it's something else | 16:30 |
| locusf | sorry, leading / missing | 16:31 |
| tbr | locusf: yeah, that looks good | 16:32 |
| tbr | your publish from irssi looks wrong though: 1407169924: Received PUBLISH from irssi_30783 (d0, q2, r0, m1, 'sailfish/locusf/irssi', ... (62 bytes)) | 16:32 |
| tbr | should add /notifications | 16:32 |
| locusf | added now | 16:33 |
| tbr | locusf: testing :) | 16:33 |
| locusf | worked | 16:33 |
| tbr | \o/ | 16:33 |
| locusf | came to my wrist too, this is hi-tech :) | 16:33 |
| tbr | I used to have that on the N950 :D | 16:34 |
| tbr | currently no metawatch client for sailfish :( | 16:34 |
| tbr | It's pretty awesome as you never have to do more than glance at your wrist to see if it's anything important on IRC | 16:35 |
| locusf | indeed, thanks for this :) | 16:35 |
| tbr | my pleasure | 16:35 |
| tbr | should put this on top of my jobsearch resume: has designed a irssi push notification system for the jolla | 16:36 |
| tbr | I'm sure that will get me a job, instantly. ;) | 16:36 |
| locusf | heheheh :) | 16:37 |
| tbr | although I applied for two ARM ltd. jobs in Oulu where they would probably appreciate "IoT protocol understanding and familiarity" | 16:37 |
| tbr | I've so far only seen _one_ client reconnect, and during that also my ssh connection to the jolla was reset. So it was probably some connman thing. | 16:45 |
| tbr | as far as irssi goes, this is now probably feature complete | 16:46 |
| tbr | unless so far I was never in full sleep and full sleep somehow also resets TCP connections. :-/ | 16:47 |
| tbr | hrm, my jolla seems to drop off the network and when it wakes up it's not connected and doesn't reconnect. although it sort of shows that it's connected. | 18:09 |
| tbr | I think it's IPv6 related | 18:10 |
| tbr | time to force an ipv6 connection to the push server | 18:11 |
| tbr | meh connman problems | 18:39 |
| tbr | only ipv6 local network connection | 18:39 |
| kimmoli | connman rulez | 19:15 |
| tbr | well, network-damager seems to have it's own share of problems with this network... | 19:16 |
| locusf | you got ipv6 local network? | 19:16 |
| tbr | was fun when I made it crash _twice_ in a day and both were critical bugs with rush fix | 19:17 |
| tbr | yes, RA based | 19:17 |
| tbr | backhaul is sadly a tunnel, as sonera sucks donkey testicles and still doesn't offer IPv6 | 19:17 |
| locusf | hmm a lot of hilights from twitter reconnect, causing irssi to deadlock | 19:28 |
| locusf | bitlbee + twitter | 19:28 |
| tbr | oh, you probably ran into the in-flight limit :( | 19:41 |
| tbr | if you want to avoid those deadlocks, the easiest for now is to set QoS to 0 | 19:41 |
| tbr | the problem is that mosquitto_pub doesn't return immediately but stalls if r=2 and it can't get the message ACKd | 19:42 |
| tbr | I was also considering to try running it as "nohup mosquitto_pub ..." but am a bit afraid to build a security hole where someone could run arbitrary shell code through hilights | 19:43 |
| tbr | I wish the MQTT for perl wasn't such a load of crap | 19:44 |
| tbr | btw: to get you guys a bit worked up: AFAICT you can put binary data in MQTT messages AND they can be quite big too | 19:51 |
| locusf | okay :/ | 19:58 |
| kimmoli | tbr i was just wondering about binary data | 20:31 |
| tbr | message size is a server side setting | 20:33 |
| kimmoli | hmm ok. just wondering to send 128x64 mono bitmap created on server, and display it on toholed | 20:33 |
| tbr | that shouldn't be a problem | 20:34 |
| kimmoli | that is just about 100 bytes | 20:34 |
| kimmoli | we noticed few days ago that imagemagick is dropped the basic bitmap header out | 20:34 |
| tbr | currently I don't have one set, so it might be 256M limit ;) | 20:35 |
| kimmoli | ok. | 20:35 |
| kimmoli | i use that BITMAPINFOHEADER in toholed /yes i made my own parser/. imagemagick now dofeault to V4, and allos to use V2 and V3 but not BITMAPINFOHEADER anymore. at least i didnt find a way (workaround DIB + add header manually) | 20:38 |
| tbr | well if you have same size/parameters every time you can just have raw bitmap and prepend the header yourself | 20:39 |
| tbr | btw: when mentioning binary capabilities I was thinking of the epaper toh, pushing things to that would be cool too. | 20:40 |
| kimmoli | yep | 20:41 |
| kimmoli | now i use shareui plugin https://github.com/kimmoli/toheink-shareui to convert image to dithered mono, and fuse https://github.com/kimmoli/epd-eink-jolla / xpd repar stuff to send it to TOH | 20:43 |
| kimmoli | image thing. nothing else done yet. | 20:43 |
| kimmoli | have to try the partial update still | 20:43 |
| kimmoli | am i missing some keys from my keyboard or wut? | 20:44 |
| tbr | either my debian install starts to suck or the keyboard on my thinkpad is wearing out. I also have missed keys etc recently. | 20:45 |
| kimmoli | or i have just used to qtcreator autocomplete :) | 20:49 |
| kimmoli | but now gn | 20:49 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!