mighty17[m] | sicelo phosh depends on normal wlroots so it automatically installs it :( | 06:41 |
---|---|---|
mighty17[m] | <buZz> "mighty17: is the flicker also..." <- Well charger isn't working in mainline xD, so this is without charger | 06:41 |
mighty17[m] | <tmlind> "PVR:(Error): PVRDisplayBufferCre..." <- Also what in the world is this error? I get it quite some times when running apps/games | 06:42 |
mighty17[m] | <tmlind> "maybe the dsi interface line..." <- Then it wouldn't show display at all? Weirdly if I use clock rate divided by 2 the display works normally | 06:43 |
mighty17[m] | <uvos> "did you dump the dss regisers in..." <- I don't remember doing that, how'd I check it on mainline? | 06:44 |
mighty17[m] | Aaaand I spammed irc with matrix replies, extremely sorry guys | 06:44 |
sicelo | mighty17[m]: phosh just needs libwlroots to be available. anything can provide it, and phosh won't know/care :-) | 08:41 |
mighty17[m] | sicelo: but phosh uses wlroots0.12 only idk why | 08:42 |
mighty17[m] | i even tried using provides="wlroots0.12" in apkbuild | 08:42 |
sicelo | right. so that's *the* problem :-) | 08:47 |
sicelo | means you need to be building that wlroots version too | 08:48 |
sicelo | or, try the wlroots 0.14 phosh - i think it's available already | 08:48 |
mighty17[m] | apk upgrade? | 08:48 |
sicelo | maybe not packaged in pmos yet. i don't know | 08:49 |
sicelo | or maybe i'm jumping the gun, but i seem to think there was some mention of a phosh branch using more recent wlroots | 08:52 |
sicelo | anyway, for now, 0.12 is clearly your friend | 08:52 |
mighty17[m] | Will stick to it | 09:06 |
uvos | freemangordon: how did the mime system work in freemantle? i noticed that the xdg mime system is broken or missing in most maemo applications | 09:52 |
freemangordon | uvos: there is libhildonmime | 11:11 |
freemangordon | ity works by using categories, which group mime type, IIUC | 11:11 |
freemangordon | *mime types | 11:11 |
uvos | looks like it dose/should use xdg mimes https://github.com/maemo-leste/libhildonmime/blob/master/libhildonmime/hildon-uri.c | 11:16 |
uvos | but those dont work for maemo packaged applicaitons | 11:16 |
uvos | (like the pdf viewer and mihphoto for starters) | 11:17 |
freemangordon | maemo applications provide additional info | 11:17 |
uvos | the .dekstop files need to provide the xdg hint | 11:17 |
freemangordon | see /usr/share/mime/packages | 11:17 |
uvos | otherwise the pdfviwer is unusable with a linux filemanager for instance | 11:17 |
freemangordon | correct | 11:17 |
freemangordon | but for maemo file manager it doesn;t matter | 11:18 |
freemangordon | at it provides some osso- stuff | 11:18 |
freemangordon | *as it | 11:18 |
uvos | well hildon-uri also uses MimeType | 11:18 |
freemangordon | buth have to be fixed though | 11:18 |
uvos | so we can unify behavior as mutch as possible | 11:18 |
freemangordon | right | 11:18 |
uvos | ok sounds good | 11:19 |
freemangordon | my concern is more about categories | 11:19 |
uvos | whats a catigory in this instance? | 11:20 |
freemangordon | we need to find a way to somehow group mime types into categories | 11:20 |
freemangordon | sec | 11:20 |
uvos | xdg dose this | 11:20 |
uvos | its image/bmp for instance | 11:20 |
uvos | image/jpeg etc | 11:20 |
freemangordon | https://github.com/maemo-leste/libhildonmime/blob/master/libhildonmime/hildon-mime.h#L40 | 11:20 |
freemangordon | no, it is a list of mime types that are of specified category | 11:21 |
freemangordon | os 'images' is gpegs, pngs, etc | 11:21 |
freemangordon | *jpegs | 11:21 |
uvos | the lib could parse this list from the desktop files it sees | 11:21 |
freemangordon | but, adobe/cdr too, for example | 11:21 |
freemangordon | it parses from packages directory | 11:21 |
uvos | same | 11:21 |
freemangordon | but, it already does :) | 11:22 |
uvos | allready dose what? | 11:22 |
uvos | parse it from the .desktop files? | 11:22 |
freemangordon | the problem is that it looks for <osso:category name="documents"/> | 11:22 |
freemangordon | no, osso-pdf-viewer.xml for example | 11:22 |
freemangordon | it is in /usr/share/mime/packages | 11:22 |
freemangordon | seems there is a bug somewhere, pdf viewver should work from modest (to open pdf attachment) | 11:23 |
uvos | so lx-open catigorizes mimes using .destop files, so it has a catigory image and then it parses all the files and if some applicaiton claimes that bmp is an image it puts the mime there | 11:23 |
uvos | this has the advantage that lx-open dosent need to know what can be an image | 11:23 |
freemangordon | smae | 11:23 |
freemangordon | *same | 11:23 |
uvos | if a new image format comes arround thats calld erf or something | 11:23 |
uvos | as sson as an app is installed what handles erf it just works | 11:24 |
freemangordon | but, do you have an example for that 'category' in .desktop file? | 11:24 |
uvos | erf as ian image category? | 11:24 |
uvos | *as an | 11:24 |
uvos | ? | 11:24 |
freemangordon | dunno | 11:24 |
uvos | yeah xdg specifies types of mimes | 11:24 |
uvos | like image/ document/ etc | 11:24 |
freemangordon | what is the key that identifies a category? | 11:24 |
uvos | MimeType=image/bmp;image/gif;image/jpeg;image/jpg;image/png;image/tiff;image/x-bmp;image/x-pcx;image/x-tga;image/x-portable-pixmap;image/x-portable-bitmap;image/x-targa | 11:25 |
freemangordon | ok, but adobe/cdr is not image neither document | 11:25 |
uvos | this is what the string looks like | 11:25 |
uvos | right | 11:25 |
uvos | some are not so descriptive | 11:26 |
uvos | or rather dont fit into those catigorys that hildon-mime uses | 11:26 |
freemangordon | exactly, that's why nokia appended to specs I guess | 11:26 |
freemangordon | no, why, adobe/cdr is image, no? | 11:26 |
freemangordon | this is example I made up, I guess there are more appropriate ones | 11:27 |
uvos | i think image is supposed to be for bitmaps only | 11:27 |
uvos | hmm no | 11:27 |
freemangordon | png is not really a simple bitmap | 11:27 |
uvos | its svg too | 11:27 |
freemangordon | right | 11:27 |
uvos | freemangordon: i meant a raster image | 11:27 |
freemangordon | ah | 11:27 |
freemangordon | either ways, it is not only that | 11:27 |
uvos | no | 11:27 |
uvos | not sure where adobe/cdr comes from | 11:28 |
freemangordon | I made it up | 11:28 |
uvos | oh ok | 11:28 |
freemangordon | I am not eve sure coreldraw still exists | 11:28 |
freemangordon | *even | 11:28 |
uvos | ok | 11:29 |
uvos | anyhow seams sane | 11:29 |
freemangordon | mhm | 11:29 |
uvos | (hildon-mime implementation) | 11:30 |
freemangordon | though, most-probably mime type would be application/cdr | 11:30 |
freemangordon | yeah | 11:30 |
freemangordon | but I would really love if we can gather the info from upstream mime types | 11:30 |
* freemangordon checks what is provided on fremantle | 11:30 | |
uvos | atho im not sure if i love the seperate xml file | 11:30 |
uvos | can you link one? | 11:31 |
uvos | not sure why its needed over just adding x-osso-something keys to .desktop | 11:31 |
freemangordon | sec | 11:31 |
freemangordon | https://pastebin.com/CAS669ct | 11:32 |
freemangordon | gimme a minute to get fremantle image viewer xml | 11:32 |
uvos | btw i dont know how osso_pdfviewer is supposed to work with mimes at all | 11:33 |
uvos | it dosent seam to allow you to open a pdf file at all via cli | 11:33 |
uvos | i tryed adding xdg mime to it beofre | 11:33 |
freemangordon | this is better example: https://pastebin.com/VjUZH8kh | 11:34 |
uvos | hmm so they have file type matching in there too | 11:35 |
freemangordon | mhm | 11:35 |
uvos | not sure we want to maintain our own database on that | 11:35 |
freemangordon | and they implement both namespaces | 11:35 |
freemangordon | me neither, that's why I am looking for a way to gather that info from upstream | 11:36 |
uvos | file --mime-type $file | 11:36 |
freemangordon | mime-type is fine | 11:36 |
freemangordon | the problem is with the category | 11:36 |
freemangordon | set | 11:36 |
freemangordon | *sec | 11:36 |
uvos | i dont think you can get the categorys you want (rather than the xdg ones) from upstream | 11:37 |
uvos | you need to match the mime-type to your own table | 11:37 |
uvos | (and use xdg ones as backup maybe) | 11:37 |
freemangordon | see https://pastebin.com/nZJMbSRv | 11:37 |
freemangordon | this is from my n900 | 11:37 |
freemangordon | and this is really useful | 11:38 |
uvos | yeah we need something like that and if we encounter a mime not in the this file | 11:38 |
uvos | we use a heuristic based on the xdg mime | 11:38 |
uvos | ie put image/ under images: and so on | 11:38 |
freemangordon | how do you know that application/pdf is a document? | 11:38 |
uvos | well we need a database we maintain for basic types | 11:39 |
uvos | and for others we hope the heuristic holds | 11:39 |
freemangordon | hmm... | 11:40 |
freemangordon | I think I have a better, albeit ugly idea | 11:40 |
uvos | try to gues by Categories= of the applicaiton that provides the mime? i dont think this can be done sufficantly realiably | 11:42 |
freemangordon | uvos: what about using generic-icon from specs? | 11:42 |
freemangordon | look at freedesktop.org.xml | 11:43 |
uvos | looks like it might hold | 11:44 |
uvos | yeah | 11:44 |
freemangordon | and we can always filter by sub-class-of | 11:44 |
uvos | yeah sounds ok ish | 11:44 |
freemangordon | mhm | 11:44 |
uvos | bit ugly ofc going by the icon name | 11:44 |
freemangordon | yeah, ugly | 11:44 |
freemangordon | but better than us maintaining our own db | 11:45 |
uvos | yeah i think so to | 11:45 |
freemangordon | ok, have to do some RL work, bbl | 11:45 |
uvos | ttyl | 11:45 |
freemangordon | will ask Wizzup and parazyd then, if they have anything to add to this | 11:46 |
freemangordon | ttyl | 11:46 |
parazyd | Why not use libmagic | 11:47 |
parazyd | Or is this about something else? | 11:48 |
uvos | parazyd: we would also use libmaic instead of the current own maic database | 11:50 |
uvos | parazyd: but this is more about catiorizing xdg mime types | 11:50 |
uvos | *magic | 11:50 |
parazyd | aha so the problem is the 'application' type mostly? | 11:51 |
uvos | yeah | 11:51 |
uvos | application/ is not descriptive enough for hildon-mimes expectations | 11:52 |
mighty17[m] | uvos: you dont mind if i upload the vid to matrix/element right? | 11:53 |
parazyd | Yeah then in case of 'application' proably sub-type needs to be matched. | 11:53 |
uvos | mighty17[m]: what about/ probubly no | 11:54 |
mighty17[m] | so where should i put it then? | 11:56 |
uvos | mighty17[m]: upload it to maxtrix and have it post a link here | 11:56 |
uvos | its fine | 11:56 |
uvos | whats the video about? | 11:56 |
mighty17[m] | yesterdays talk | 11:57 |
uvos | ok lets see it | 11:57 |
mighty17[m] | about pvr artifact | 11:57 |
mighty17[m] | uvos: data:video/mp4, | 12:08 |
uvos | :( | 12:08 |
uvos | i would have expected matrix to post a link | 12:08 |
Wizzup | we just see a mimetype | 12:08 |
uvos | mighty17[m]: cant see it | 12:08 |
mighty17[m] | https://matrix.to/#/!xbgJyIUhPvTgUGSPDF:matrix.org/$2deJ6vWCP8T8ibnYMNbV8cX10doKy0T_UC0ZZavNzyE?via=matrix.org | 12:08 |
Wizzup | requires an app to be installed | 12:08 |
mighty17[m] | ughh | 12:09 |
mighty17[m] | gimme a sec | 12:09 |
* mighty17[m] uploaded a video: (6622KiB) < https://libera.ems.host/_matrix/media/r0/download/matrix.org/aXANFphNFWzxriQATLKPiEUT/VID_20210930_153955.mp4 > | 12:10 | |
uvos | works | 12:11 |
uvos | mighty17[m]: hmm ok i dont see any artifacting but on d4 its only visible/noticeable durring smooth animations | 12:11 |
mighty17[m] | uvos: maybe coz it goes from 0-60hz? | 12:13 |
uvos | yeah maybe because of that | 12:13 |
uvos | but on d4 if you display over hdmi it also occures | 12:13 |
uvos | (where its locked to 60) | 12:14 |
uvos | i would be interesting if your device dident do it at all | 12:14 |
mighty17[m] | i dont have hdmi support yet so cant test that | 12:14 |
uvos | since we could try and find the differenc | 12:14 |
mighty17[m] | i still think its due to my refresh rate being locked at 30hz | 12:14 |
uvos | but im not sure that video is conclusive again it only happens in smooth 3d accelerated operations | 12:15 |
* mighty17[m] uploaded a video: (7507KiB) < https://libera.ems.host/_matrix/media/r0/download/matrix.org/DsQnwMjBBPekceVpNVfedBQp/VID_20210930_152753.mp4 > | 12:15 | |
mighty17[m] | heres it at 60hz | 12:15 |
mighty17[m] | smooth 3d accelerated operations, phosh is 3d accl :P | 12:15 |
mighty17[m] | what do you want me to try tho | 12:15 |
uvos | hildon-desktop idealy :P | 12:17 |
mighty17[m] | well not possible on pmOS :P | 12:17 |
uvos | i think the mesa demo es2_tri showed this problem | 12:17 |
uvos | if you clicked space offent enough | 12:17 |
uvos | but i think it only works on x | 12:17 |
mighty17[m] | no x on 1.17 :( | 12:18 |
uvos | hmm yeah | 12:19 |
uvos | have you run plamo on it before? | 12:19 |
mighty17[m] | plamo fails are pvr is missing some egl config | 12:20 |
uvos | ok | 12:20 |
uvos | hmm | 12:20 |
uvos | then idk | 12:20 |
mighty17[m] | sicelo: managed to run it once | 12:20 |
uvos | on d4 yeah | 12:20 |
mighty17[m] | glmark or smth? | 12:20 |
uvos | idk if it shows in glmark | 12:20 |
uvos | maybe | 12:20 |
uvos | try it and watch carefully | 12:20 |
mighty17[m] | well i'll ask tell if it crashes pvr | 12:21 |
mighty17[m] | s/ask/also | 12:21 |
Wizzup | mighty17[m]: maybe add the things necessary for leste port? | 12:22 |
mighty17[m] | Wizzup: leste already used to run with d4 rootfs but then my sdcard reader broke :( | 12:23 |
uvos | :\ | 12:23 |
Wizzup | mighty17[m]: still, would be nice to get the kernel and other changes somewhere | 12:23 |
mighty17[m] | guess what! pvr crashed https://paste.debian.net/1213842/ | 12:24 |
mighty17[m] | Wizzup: its in pmOS | 12:24 |
uvos | mighty17[m]: hmm ok | 12:24 |
uvos | mighty17[m]: no idea why your pvr is so crashy | 12:24 |
mighty17[m] | i can send a patch here to see my dts if needed | 12:26 |
uvos | not that usefull to us | 12:26 |
* Wizzup has like 7 usb sd card readers | 12:26 | |
uvos | just continue improving your device and we can add it to our kernel when its ready | 12:27 |
mighty17[m] | https://paste.debian.net/1213844/ <- dts and https://pastebin.com/tmxvkkPE <- config, maybe i have something set wrong | 12:29 |
uvos | mighty17[m]: so besides the dsi problems whats the general working state of the deivce rn? | 12:29 |
mighty17[m] | uvos: well i'd like to fix 60hz and the backlight, but idkk what to do | 12:29 |
mighty17[m] | uvos: sound and light sensor dont work (ie i havent started to work on them yet), cameras dont (thanks to ducati) well everything else works | 12:30 |
uvos | with the 60hz so i would grab the omap4 datasheet and dump the registers related to the dss | 12:30 |
uvos | and do the same on android | 12:30 |
uvos | and see if they are different | 12:30 |
uvos | and then try and parse the difference | 12:30 |
uvos | use rwmem its pretty easy to build with the ndk for android | 12:31 |
uvos | no idea about the backlight | 12:31 |
uvos | btw if you want we can send you a d4 to mess around with as well | 12:32 |
uvos | might be usefull to compear stuff and also maybe you can help us improve omap4 support in general | 12:32 |
mighty17[m] | nah i dont think im the best guy to send a d4 to, d4s are already very rare :P | 12:33 |
uvos | not that rare but as you wish | 12:33 |
mighty17[m] | uvos: im kinda sure i have the wrong pwm value, but cant find the correcet one | 12:33 |
uvos | its connected to a omap4 pwm? | 12:34 |
mighty17[m] | also did you see the 60hz vid i sent, maybe that'll help debugging (if its like a known issue or smth) | 12:34 |
mighty17[m] | uvos: gptimer | 12:34 |
tmlind | mighty17[m]: probably good idea to test also on d4 so you can compare | 12:34 |
mighty17[m] | d4 and tab2 are quite different | 12:35 |
Wizzup | freemangordon: any idea about fremantle and firewall rules? | 13:21 |
bencoh | Wizzup: nothing, according to iptables | 13:23 |
bencoh | ie policy ACCEPT with no restriction whatsoever | 13:23 |
mighty17[m] | Or maybe uvos it's a 1.9 thing (the artifacting) | 13:23 |
bencoh | (which is kinda scary tbh) | 13:23 |
bencoh | I'm not certain it should go into anything "leste" anyway | 13:24 |
uvos | mighty17[m]: no it happens on d4 on ddk 1.17 | 13:26 |
mighty17[m] | Hildon works with 1.17 now? That's news to me | 13:28 |
uvos | no | 13:28 |
Wizzup | bencoh: sorry, not certain what should go into anythign leste? | 13:28 |
uvos | but it do also have d4's that run debian/sway | 13:28 |
bencoh | Wizzup: firewall management | 13:28 |
Wizzup | my mobile isp still gives me public v4s on my phone | 13:28 |
uvos | on a phone its sane to discard all incomeing by default or? | 13:28 |
bencoh | I'm not certain leste should take care of that part | 13:28 |
Wizzup | + stuff like tor transproxy requires iptable setups | 13:29 |
bencoh | Wizzup: seriously? neat | 13:29 |
Wizzup | just looking into nftables vs iptables | 13:29 |
Wizzup | bencoh: neat in some way, not really for my poor old linux fremantle n900 | 13:29 |
bencoh | Wizzup: I use iptables on fremantle | 13:29 |
Wizzup | debian buster+ seems to suggest nftables | 13:29 |
bencoh | you can just drop any incoming connection | 13:30 |
bencoh | oh, they finally moved to nftables? interesting | 13:30 |
uvos | iptables is mostly maintinaince mode now i would avoid using it | 13:30 |
uvos | all desktop distros have swiched afaik | 13:30 |
bencoh | well, I guess I still live in the past ... time to learn nftables | 13:30 |
bencoh | I'll probably hate it if it's a pf-like syntax though | 13:31 |
uvos | no idea my router pc is still on iptables | 13:31 |
uvos | so yeah im also lazy | 13:31 |
bencoh | I first heard about nftables 13 years ago, so ... it was about time I guess :D | 13:32 |
freemangordon | is that kernel-side scripting? | 13:35 |
freemangordon | Wizzup: I am with bencoh on that - it is not us to deal with FW rules | 13:36 |
freemangordon | we *may* provide a UI someday for user to have some control | 13:36 |
bencoh | they already replaced the kernel-side iptables/ebtables/whatever code with a common nft engine, and wrote a compatibility layer for the userspace ip/eb/arp/tables tools; but now users should migrate to the native userspace nft tool | 13:37 |
Wizzup | freemangordon: really? imho that's insane | 13:38 |
Wizzup | but ok | 13:38 |
bencoh | freemangordon: Wizzup might have a point regarding tor applets and the likes though | 13:38 |
bencoh | but I'm not certain we can provide a good solution | 13:38 |
Wizzup | I think the sane default is to *block* all stuff but ssh wrt open ports, there is no reason for a mobile device to just have all ports open | 13:38 |
uvos | i also think since haveing a public ip on cellular | 13:39 |
Wizzup | but we can also just use default nft rules in debian which allow everything | 13:39 |
uvos | its not sane to accept everything | 13:39 |
Wizzup | yeah | 13:39 |
uvos | i would at least have a rule that rejects all incomeing on wwan* | 13:39 |
freemangordon | but there should be a way (UI) for the user to punch through the FW | 13:40 |
Wizzup | bencoh: there are iptables binaries that just work | 13:40 |
bencoh | Wizzup: where/what? | 13:40 |
Wizzup | bencoh: as in they can work on nftables | 13:40 |
Wizzup | so there's a lot of compatibility there | 13:41 |
bencoh | ah, yeah, that's pretty much what I tried to explain | 13:41 |
bencoh | but I guess that wasn't clear :) | 13:41 |
Wizzup | ok | 13:41 |
uvos | if you just reject everything on wwan i think its fine without a ui | 13:42 |
uvos | someone accessing the phone via its public ip on the cellular network is nice enough | 13:42 |
bencoh | uvos: if you really do that you should allow disabling that part | 13:42 |
freemangordon | :nod: | 13:42 |
Wizzup | freemangordon: sure there will be ways to configure a firewall, there are many | 13:42 |
uvos | that i think its sane to have the user use cli | 13:42 |
freemangordon | no | 13:42 |
uvos | that nice enouth | 13:42 |
uvos | niche | 13:42 |
freemangordon | this is not desktop | 13:42 |
Wizzup | we could use https://firewalld.org/ | 13:42 |
uvos | ? | 13:42 |
uvos | right its not desktop | 13:42 |
freemangordon | phone is not a desktop | 13:43 |
uvos | accessing the device via incomeing conecction on cellular is so out there | 13:43 |
bencoh | the thing is ... once you start fiddling with firewall rules, you will always get in the way of power users | 13:43 |
uvos | its not "phone" usage at all | 13:43 |
Wizzup | the point is that the only sane default is to reject all but the necessary input traffic (e.g. ssh maybe) | 13:43 |
Wizzup | we don't have to do it now, but eventually that will be a default | 13:43 |
freemangordon | and why not torrents? or RTP? | 13:44 |
Wizzup | it's pretty easy to add more ports to exclusion via whatever ui/ux you want | 13:44 |
Wizzup | torrents can work fine through firewalls | 13:44 |
bencoh | uvos: at some point I used tinc on my phone as well to access from anywhere | 13:44 |
Wizzup | afaik | 13:44 |
freemangordon | I want nothing basically in that regard :) | 13:44 |
bencoh | Wizzup: it's messy as well | 13:44 |
uvos | bencoh: and you have no poblem configuring the firewall yourself | 13:44 |
uvos | bencoh: people that do this sort of thing are a very select crowd | 13:44 |
freemangordon | there is no active FW on fremantle | 13:44 |
bencoh | uvos: sure, but then I don't want my mobile-ish subsytem to push rules here and there | 13:45 |
freemangordon | exactly | 13:45 |
freemangordon | ny moint | 13:45 |
freemangordon | *my | 13:45 |
uvos | sure but a config file is suffichent here | 13:45 |
uvos | no ui nessecary | 13:45 |
uvos | or cli command | 13:45 |
freemangordon | as soon as there is no default "restrict everything" | 13:45 |
freemangordon | as soon as there is such a rule, you should be able to remove it through UI | 13:46 |
freemangordon | GUI even | 13:46 |
bencoh | "cli command" probably means trouble, config file might be enough, but then you'll have to make sure you don't break stuff (torrent / SIP / RTP / whatever) | 13:46 |
bencoh | and since I guess we'll always break something, fmg is probably right | 13:46 |
bencoh | a simple checkbox in the network/connection panel would work btw | 13:47 |
freemangordon | mhm | 13:47 |
Wizzup | I strongly disagree but also don't care to discuss it more atm | 13:47 |
freemangordon | ok, lets discuss when it comes to it :) | 13:47 |
bencoh | alright :) | 13:47 |
freemangordon | I also care more about mime types ATM | 13:48 |
freemangordon | so, back to | 13:48 |
freemangordon | uvos: I guess x-office-presentation is document, not image, what do you think? | 15:52 |
Wizzup | sounds like power point slides | 15:53 |
Wizzup | probably provided by libreoffice | 15:53 |
freemangordon | sure, the point is which category to put that | 15:53 |
freemangordon | 'documents' or 'images' | 15:53 |
freemangordon | Wizzup: in terms of libhildonmime that is | 15:54 |
Wizzup | documents | 15:54 |
freemangordon | https://github.com/maemo-leste/libhildonmime/blob/master/libhildonmime/hildon-mime.h#L40 | 15:54 |
freemangordon | ok | 15:54 |
Wizzup | meanwhile wg works nicely for public v4 - my d4 = 95.217.97.174 | 15:55 |
Wizzup | (just for now of course) | 15:56 |
Wizzup | not even one attempt in auth.log I am disappointed ;) | 15:58 |
Wizzup | turned it off again | 15:58 |
sicelo | :-) | 15:59 |
freemangordon | uvos: images: application/illustrator application/vnd.corel-draw :p | 16:20 |
freemangordon | this is with algo I just came up with | 16:20 |
freemangordon | it is a bit smarter than just looking at the icons though | 16:20 |
parazyd | freemangordon: What widget is it in libhildon we can use to browse and select a file on the system? | 16:23 |
freemangordon | it is in hildon-fm | 16:24 |
freemangordon | HildonFileChooser or somesuch | 16:24 |
uvos | freemangordon: what algo? | 16:24 |
freemangordon | will push in a while, but basically: | 16:24 |
uvos | freemangordon: why is there a seperate HildonFileChooser instead of the gtk file chooser simply being changed to be touch frendly? | 16:25 |
freemangordon | because it has almost nothing in common with gtk file chooser | 16:25 |
uvos | well you can just replace the code in qt and gtk3 you can also have the file choose live in a plugin | 16:25 |
freemangordon | 1. if media type is image audio etc, it assigns the appropriate category, otherwise | 16:27 |
freemangordon | 2. if there is no category assigned by the type, but there is icon, assign category based on the icon (if applicable) | 16:27 |
freemangordon | 3. if there is also osso category provided, use it if there is no category assigned by 1 and 2 | 16:27 |
uvos | ok | 16:27 |
uvos | sounds sane | 16:27 |
freemangordon | if both 1 or 2 and osso assign category, prefer the one assigned by 1 or 2 | 16:27 |
uvos | ok | 16:28 |
uvos | did you replace the magic in the xml files with something else (libmagic)? | 16:28 |
freemangordon | this https://pastebin.com/7aBHV9nuis the result, no idea how to verify if it is correct | 16:28 |
freemangordon | the code I deal with does not touch those | 16:29 |
uvos | wrong link | 16:29 |
freemangordon | hmm works for me | 16:29 |
uvos | This page is no longer available. It has either expired, been removed by its creator, or removed by one of the Pastebin staff. | 16:29 |
freemangordon | oh, sorry | 16:29 |
freemangordon | yeah | 16:29 |
freemangordon | https://pastebin.com/7aBHV9nu | 16:30 |
uvos | documents | 16:30 |
uvos | application/mathematica | 16:30 |
uvos | hmm | 16:30 |
freemangordon | hmm? | 16:30 |
freemangordon | is that wrong? | 16:30 |
uvos | its a script | 16:30 |
freemangordon | well, doc file is a script too, no? | 16:31 |
freemangordon | .doc | 16:31 |
freemangordon | more or less | 16:31 |
uvos | sortof yeah | 16:31 |
uvos | but mathematica files are more so scripts | 16:31 |
uvos | but i gues its arguable | 16:31 |
uvos | it looks fine | 16:33 |
freemangordon | <generic-icon name="x-office-document"/> | 16:33 |
uvos | im sure its its mostly correct enough to not get in peoples way | 16:33 |
freemangordon | this is for application/mathematica | 16:33 |
freemangordon | mhm | 16:33 |
freemangordon | I don;t think we can do it any better | 16:33 |
uvos | if both 1 or 2 and osso assign category, prefer the one assigned by 1 or 2 | 16:35 |
uvos | maybe you want to flip this | 16:35 |
uvos | and remove all entrys | 16:35 |
uvos | so we can selective override later | 16:35 |
uvos | if we find something thats annoying | 16:35 |
freemangordon | not sure I understand correctly | 16:35 |
freemangordon | ah | 16:36 |
freemangordon | hmm | 16:36 |
freemangordon | right | 16:36 |
freemangordon | ok | 16:36 |
freemangordon | uvos: wait, can;t be done | 16:37 |
freemangordon | because files are parsed in arbitrary order | 16:37 |
freemangordon | actually, the check makes sense only if we have both freedesktop and osso in the same file | 16:38 |
uvos | ah ok | 16:38 |
freemangordon | otherwise whichever comes first remains, I refuse to add the same category more then once | 16:39 |
freemangordon | *the same type | 16:39 |
uvos | well then scratch that, i gues the code could be improved by first parsing all files to grab the mimes and then assigning the categorys | 16:39 |
uvos | but its fine as is if thats too mutch work | 16:39 |
freemangordon | mhm | 16:39 |
freemangordon | sure it can be done, but lets do it when we really need it | 16:40 |
freemangordon | for now I think wat we have is fine | 16:40 |
uvos | yes | 16:40 |
Wizzup | freemangordon: bencoh: looks like we have dns port opened on leste by default (dnsmasq probably binds to all interfaces) | 16:48 |
Wizzup | seems like a firewall would easily take care of that problem | 16:48 |
buZz | imho, just bind it to localhost | 16:49 |
Wizzup | sure, that too | 16:49 |
buZz | dont have to firewall :P | 16:49 |
bencoh | Wizzup: :) | 16:49 |
Wizzup | defense in depth is better | 16:49 |
buZz | hmhm | 16:49 |
bencoh | tbh I don't understand the default dnsmaq behavior on debian* | 16:50 |
mighty17[m] | uvos: regarding PlaMo here's my log https://pastebin.com/ZvCSdQy2 any clues? | 16:50 |
Wizzup | most debian default behaviour I don't understand :P | 16:50 |
bencoh | haha, same :) | 16:50 |
Wizzup | latest libicd-network-wireguard and libicd-network-tor should support proxying everything (for wireguard, set allowedips to 0.0.0.0/0, for tor, just enable transparent proxying) | 16:52 |
Wizzup | the tor stuff is still building fwiw | 16:52 |
Wizzup | parazyd: we need to fix the network_modules writing | 17:01 |
Wizzup | it happens on my d4 now too | 17:01 |
parazyd | Yep | 17:01 |
freemangordon | hmm, now the icon is wrong :( | 17:04 |
Wizzup | hehe... wireguard tunnel system wide and tor system wide transproxy work together | 17:08 |
Wizzup | neat. | 17:08 |
kona | Freemangordon: Corel desperately believes that Draw still exists. https://www.coreldraw.com/en/ | 17:12 |
Wizzup | https://wizzup.org/surf-tor.png | 17:12 |
freemangordon | kona: mhm | 17:18 |
freemangordon | yay, I see thumbnails when I want to attach image file :D | 17:18 |
Wizzup | :D | 17:19 |
freemangordon | parazyd: could you add hildon-thumbnail to the appropriate meta-package? | 17:19 |
freemangordon | without that package no image thumbnails are shown when you browse the filesystem | 17:20 |
freemangordon | ok, this starts to look and feel like maemo :D | 17:23 |
Wizzup | :) | 17:26 |
bencoh | <3 | 17:28 |
bencoh | (not that I was really happy with the tracker-indexer / thumbnailer though :]) | 17:29 |
Wizzup | hehe | 17:29 |
freemangordon | image thumbnailer is a dbus service which gets called when needed, IIUC | 17:35 |
freemangordon | and we still do not have a tracker | 17:36 |
Wizzup | freemangordon: I see this on upgrade: | 17:36 |
Wizzup | * Error in type 'application/x-stellarium-script' | 17:36 |
Wizzup | * (in /usr/share/mime/packages/stellarium.xml): | 17:36 |
Wizzup | * Unknown freedesktop.org field 'icon'. | 17:36 |
Wizzup | *** | 17:36 |
Wizzup | not sure if it's a problem | 17:36 |
freemangordon | who gives that error? | 17:37 |
Wizzup | Setting up hildon-update-category-database (3.1.0+2m7) ... | 17:37 |
bencoh | stellarium, as in, the astronomy thing? | 17:38 |
freemangordon | I guess hildon-update-category-database does not understabd that element | 17:40 |
freemangordon | do you have categories file created in usr/share/mime? | 17:40 |
freemangordon | Wizzup: ^^^ | 17:40 |
Wizzup | freemangordon: do you ask if this file exists? /usr/share/mime/icons | 17:41 |
Wizzup | # cat /usr/share/mime/icons | 17:41 |
Wizzup | application/x-stellarium-script:stellarium | 17:41 |
freemangordon | no /usr/share/mime/categories | 17:42 |
Wizzup | yes | 17:42 |
freemangordon | does it look sane? | 17:42 |
Wizzup | I think so, I didn't follow along much, it contains 8 lines, some really long | 17:42 |
freemangordon | right | 17:42 |
freemangordon | it is fine then | 17:42 |
freemangordon | we may want to teach hildon-update-category-database to ignore "icon" instead of error out | 17:43 |
freemangordon | but it looks harmless | 17:43 |
freemangordon | Wizzup: here https://github.com/maemo-leste/libhildonmime/blob/master/libhildonmime/update-category-database.c#L306 | 17:44 |
freemangordon | have to run, bbl | 17:44 |
Wizzup | ttyl | 17:48 |
uvos | freemangordon: btw we also need to unify X-Osso-URI-Scheme and desktop linux x-scheme-handler | 18:22 |
uvos | eg xdg-open and hildon-uri should choose the same thing for sms://029300902938903 | 18:22 |
uvos | and http:// and so on | 18:23 |
freemangordon | uvos: :nod: | 21:26 |
Wizzup | freemangordon: very fun to hear my d4 vibrate every now and then and actually show new mails | 21:41 |
Wizzup | also with the blue led indicator | 21:41 |
freemangordon | toldya, it starts to look and feel like maemo :) | 21:43 |
freemangordon | Wizzup: parazyd: btw, we want iphbd and iphb-dkms along with thumbnailer | 21:44 |
Wizzup | dkms? | 21:44 |
freemangordon | mhm | 21:44 |
freemangordon | kernel module | 21:44 |
Wizzup | what does it do? | 21:44 |
freemangordon | provides kernel support for iphbd | 21:44 |
freemangordon | https://github.com/maemo-leste/iphb-dkms | 21:44 |
Wizzup | what does iphbd do | 21:44 |
freemangordon | ip heartbeat | 21:45 |
freemangordon | daemon | 21:45 |
freemangordon | groups ip traffic | 21:45 |
freemangordon | decreases radio wakeups | 21:45 |
Wizzup | hmm | 21:45 |
Wizzup | is this something we can do with nftables or other modern kernel modules? | 21:45 |
freemangordon | I don;t think so | 21:45 |
Wizzup | it sounds like traffic shaping | 21:45 |
freemangordon | sounds like, but is not | 21:46 |
freemangordon | https://github.com/maemo-leste/iphbd/blob/master/iphbd/libiphb.h | 21:47 |
freemangordon | it has userspace API as well | 21:47 |
Wizzup | ok | 21:47 |
Wizzup | well, I would be of the opinion that we can leave it out for now, but if you want to work on it, sure we can include it | 21:48 |
freemangordon | https://talk.maemo.org/showthread.php?t=89709 | 21:48 |
freemangordon | what do you mean? | 21:48 |
freemangordon | we already have it, it just needs to be included in the metapackage | 21:48 |
Wizzup | ah, ok | 21:49 |
freemangordon | modest uses it too BTW | 21:49 |
Wizzup | yeah the thread you linked mentioned it | 21:50 |
freemangordon | look at this https://github.com/maemo-leste/iphbd/blob/master/iphbd/hbtest.c#L6 | 21:51 |
freemangordon | so, it kinda 'shapes' the traffic, but allows every application to set its own 'wake' timeout | 21:52 |
freemangordon | and there is logic that tries to wake-up all the requesting applications at the same moment, vastly saving battery (in theory at least) | 21:53 |
freemangordon | oh, seems parazyd added thumbnailer to the metapackage, thanks! | 21:55 |
parazyd | Didn't build it yet | 21:55 |
freemangordon | yeah | 21:55 |
parazyd | But feel free to trigger whenever | 21:55 |
freemangordon | maybe add iphbd and iphb-dkms firsrt | 21:56 |
freemangordon | though, lemme try if it builds still | 21:56 |
parazyd | Is it tested on devices? | 21:56 |
parazyd | Yeah | 21:56 |
freemangordon | last time I tried was on kenel 4.6 or something | 21:56 |
freemangordon | parazyd: no, but it is not device dependent | 21:56 |
freemangordon | at least for 4.19.0-17-amd64 it is fine | 22:02 |
freemangordon | parazyd: but, how to add dependency to kernel headers? | 22:03 |
Wizzup | I should go on talk.maemo.org some more | 22:03 |
freemangordon | hmm? | 22:03 |
parazyd | freemangordon: "linux-headers" I think | 22:03 |
parazyd | It's a virtual pkg | 22:03 |
freemangordon | Wizzup: not that it is bad idea, but why in particular? | 22:04 |
freemangordon | parazyd: ok, will try | 22:04 |
freemangordon | E: Package 'linux-headers' has no installation candidate | 22:04 |
parazyd | Then "linux-headers-amd64 | linux-headers" | 22:04 |
parazyd | I suppose it doesn't exist on the VM | 22:05 |
parazyd | But we have it on d4 kernel for example. | 22:05 |
freemangordon | and what about on the devices? | 22:05 |
uvos | i would strongly reccomend against something that needs out of tree kernel modules | 22:05 |
parazyd | freemangordon: https://github.com/maemo-leste/droid4-linux/blob/maemo/beowulf-devel/debian/control#L27 | 22:05 |
freemangordon | uvos: why? | 22:05 |
freemangordon | since when the kernel is a holy cow we should not touch? | 22:06 |
uvos | because we need manny kernels at different versions for different devices and maintaing that is going to be a nightmare | 22:06 |
freemangordon | n | 22:07 |
freemangordon | no | 22:07 |
uvos | also we want stuff to just work on $random device | 22:07 |
freemangordon | it is 50 LOC module | 22:07 |
freemangordon | and it will work | 22:07 |
uvos | i dont care if its 50 lok or 500000000000000000000000000000000000000000000000000000000000 lok | 22:07 |
uvos | it a terrible idea | 22:07 |
Wizzup | I mostly think it's not super relevant right now | 22:07 |
freemangordon | well, unless you provide a better way to have the same functionality, this is what we have | 22:07 |
uvos | laso its a 600 lok module | 22:11 |
freemangordon | yes, 600 | 22:12 |
freemangordon | so? | 22:12 |
freemangordon | it's been supported for the last 10 years, that would give you a clue of how hard is it to maintain | 22:12 |
freemangordon | it originated from linux 2.6.28 | 22:13 |
uvos | it also dosent really do what you say | 22:13 |
freemangordon | what did I say about the kernel module? | 22:14 |
uvos | it dosent group ip traffic in any meainfull sense of the word | 22:14 |
freemangordon | this is what iphnd does | 22:14 |
freemangordon | by waking-up the registered applications | 22:14 |
uvos | just delay tcp keepalive pakets specificly and hope that more come in | 22:14 |
freemangordon | kernel module groups keep-alives | 22:14 |
freemangordon | yes, it groups them ,see the flush code https://github.com/maemo-leste/iphb-dkms/blob/master/iphb.c#L151 | 22:16 |
Wizzup | uvos: do you know if it is possible to mirror the d4 screen over hdmi | 22:34 |
Wizzup | (for capture purposes) | 22:34 |
uvos | yes | 22:34 |
uvos | works fine | 22:34 |
uvos | why? | 22:35 |
uvos | xrandr --output HDMI-1 --auto --output DSI-1 --off ; killall hildon-desktop | 22:36 |
uvos | i have a script called /usr/bin/switch-to-ext-dpl | 22:36 |
uvos | that dose that | 22:36 |
uvos | mirror also works fine | 22:36 |
uvos | just remove the --output DSI-1 --off | 22:36 |
uvos | but then the external display needs to support the same resolution | 22:37 |
uvos | otherwise hildon will freak out | 22:37 |
uvos | hildon freaks out anyways | 22:37 |
uvos | thats why it needs to be restarted with killall | 22:37 |
uvos | hdmi audio dosent work | 22:38 |
uvos | i think it works again in 5.14 | 22:38 |
uvos | freemangordon: so im looking at the kernel src, havent found it yet | 22:39 |
uvos | but im quite sure that the kernel uses a low accuracy qos timer for keepalives | 22:39 |
uvos | so the module is quite pointless | 22:39 |
uvos | i would also view iphbd in userspace with extream suspican | 22:40 |
uvos | but have other stuff to do rn | 22:40 |
Wizzup | ok @ mirror | 22:48 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!