Wikiwide_ has quit [Remote host closed the connection]
Wikiwide_ has joined #maemo-leste
Wikiwide_ is now known as Wikiwide
joerg has quit [Ping timeout: 260 seconds]
joerg has joined #maemo-leste
mardy has joined #maemo-leste
freemangordon has joined #maemo-leste
Wikiwide has quit [Read error: Connection reset by peer]
elastic_dog has quit [Ping timeout: 250 seconds]
elastic_dog has joined #maemo-leste
peetah_ has quit [Ping timeout: 260 seconds]
peetah has joined #maemo-leste
<sicelo>
missMyN900: and i never had a problem with wpa_supplicant ever, even on N900. whether using manual methods (wpa_cli) or frontends. anyway, no denying that everyone is moving towards iwd, and one may as well embrace it
Pali has joined #maemo-leste
ruleh has joined #maemo-leste
uvos has joined #maemo-leste
<uvos>
parazyd: can i have a extras repo for qtwebbrowser?
<uvos>
i dislike it because it forces ofono on devices with no modem
<crab>
cool...
<Wizzup>
uvos: by design it system waits for sim to be ready before it continues
<Wizzup>
this is how fremantle does it
<Wizzup>
maybe it can not block if ofono is not running
<Wizzup>
or something.
<uvos>
Wizzup: im pretty sure you can just after: ofono and have the script return of ofono is not running
<Wizzup>
it's not about the script, it's about the binaries it invokes
<Wizzup>
and in general this contains a lot of non-ofono stuff AFAIK
<uvos>
well fixing the binary would be best
<Wizzup>
so it can't just *not* run
<uvos>
what is it?
<uvos>
the binary
<Wizzup>
sure if we remove the ofono line now it will work for non-devel, but anyone with ofono and the connui-cellular installed will get a black screen forever
<Wizzup>
startup-pin-entry iirc
<Wizzup>
it's supposed to block in X before h-d starts
<uvos>
and what dose startup-pin-entry do except unlock the sim?
<Wizzup>
^
elastic_dog has joined #maemo-leste
<Wizzup>
there's a whole part that uses actual matchbox
<uvos>
yeah but what for? why would we block x on a device with no modem?
<uvos>
for what
<Wizzup>
nobody said 'with no modem'
<Wizzup>
of course with devices without modem it is different
<uvos>
ok well why would we block x on a device with ofno ot installed?
<Wizzup>
I didn't say we should block X if there is no modem waiting for pin
<uvos>
ok
<crab>
hopefully i did that right
<crab>
the copy paste was a bit tricky
<uvos>
what i realy want to know is why must startup-pin-entry run if ofono is not availble
<crab>
but since it thinks those things exist i guess it cant have been that badly wrong
<uvos>
run as in more than just exit with: ofono not available
<Wizzup>
uvos: I don't think it should be installed when there is no modem probably
<Wizzup>
that's an easy solution, but it doesn't help if we then remove 'ofono' from the depends of this script
<uvos>
why you can have it after: ofono
<uvos>
then iiuc if ofono isten installed
<uvos>
the script will still work fine
<uvos>
while if it is ofono will run before this script
<Wizzup>
crab: so I don't understand this, I see parazyd committed those packages at some point but I reverted them, but it is not clar to me what made it into the repo
<Wizzup>
uvos: ok, I would try it on my d4 except that ofono there isn't helpful :P
<uvos>
btw startup-pin-entry needs to be replaced anyhow
<uvos>
since sim cards can be hot plugged etc
<Wizzup>
it can run upon sim plug as well then
<uvos>
right
<Wizzup>
in any case ofono often doesn't detect this currently
<uvos>
we can change it from blocking the session
<Wizzup>
I wouldn't worry about this edge case atm
<uvos>
to just poping up whenever ofono reports a new sim
<Wizzup>
uvos: yeah it only blocks the session by 'virtue' of it not exiting and not being backgrounded
<Wizzup>
so I should test this with 'after' and see if I still get the pin entry?
<uvos>
not sure how you would test that, i dont get pin entry regardless
<Wizzup>
wish the droid4 and n900 sim size were the same hehe
<uvos>
due to ofono isues
<uvos>
oh on n900 yeah
<Wizzup>
oh, I forgot I removed nokia-modem from 5.15 because of panics
<Wizzup>
rip
<Wizzup>
SNAFU all day long :P
<crab>
Wizzup: so essentially on my machine, apt wants those two packages in the repo for the gui (or something it depends on - the wifi thing) ?
<Wizzup>
crab: let me check something with parazyd
<uvos>
Wizzup: maybe making startup-pin-entry non blocking and wait for ofono to show on dbus to pop up with a window is least work
<uvos>
then the init script dosent have to care about ofono at all
<Wizzup>
uvos: well that's not how it should work I think
<uvos>
why not
<uvos>
you want the phone to be unusable with a locked sim?
<uvos>
as in to someone without the pin
<uvos>
i allways thought that was a absurd security mesure.
<uvos>
since the attecker can just swap the sim, or remove it
<crab>
Wizzup: i appreciate this, but remember: im just a casual hobbyist end user who is (hopefully) helping you lot out. my n900 is doing what it needs to do and as ive said before i am more than happy to do a reinstall at some point if ive been unlucky :)
<Wizzup>
uvos: I don't know why they do it per se, but I think it's a logical order
<Wizzup>
crab: right, do you still have ui problems?
<crab>
oh sure
<uvos>
i dont think so, but ok
<Wizzup>
can you change one line and see if it gets better
<crab>
i have no ui at all except during kernel boot :)
<crab>
sure ill try...
<crab>
(well until i ssh in ofc)
<crab>
what do you need me to change?
<Wizzup>
crab: edit /etc/init.d/xsession and remove 'ofono' from it
<crab>
roger...
<Wizzup>
and then restart/reboot
<crab>
roger again...
<Wizzup>
the wlan thing we'll fix next, it looks like the CI just failed
* crab
waves at n900
n900 has quit [Ping timeout: 250 seconds]
<uvos>
Wizzup: looks like the right flag is use: not after:
<freemangordon>
uvos: the other option is to have custom dh_something in debian/$notsurethepath that installs .desktop file
<freemangordon>
but I have never used that so not sure how it works
<Wizzup>
freemangordon: hm, I do occasionally get some segfaults still with cma=32MB: [ 2402.701477] cma: cma_alloc: reserved: alloc failed, req-size: 375 pages, ret: -16
<freemangordon>
Wizzup: -depth 16 is your friend ;)
<uvos>
on n900 depth 16 might acutally be worthi it
ruleh has quit [Quit: Client closed]
<uvos>
it cant run the applications i have in mind that dont do 16 bit anymore
<uvos>
anyways
<freemangordon>
until I implement that patch that scanouts should not require CMA
<freemangordon>
and even after then, it doesn;t make sense to have 24bpp on mobiles with 3-4 inches display
<freemangordon>
uvos: what are those apps?
<enyc>
crab: I'm here too but not the same as you =)).
<crab>
hahah
<crab>
i bothered to check if sicelo was in the other other place but i never thought you'd be in here too!
<Wizzup>
freemangordon: do you use it with 16 bit depth atm?
uvos has quit [Remote host closed the connection]
<sicelo>
crab: i just remembered that enyc was the other person who reported ipv6 issues, hence my question
<crab>
oh!
<crab>
yeah enyc and i love the ipv6
<crab>
we cant get enough of it
<Wizzup>
freemangordon: if so, please let me know what config you use and I'll try it
<Wizzup>
otherwise, I think I will increase cma size for now to make this usable
<crab>
sicelo: it was just a bit freaky after the bauhaus thing! :)
<crab>
i guess its a small world...
<freemangordon>
Wizzup: I dont; use it at all :)
<enyc>
sicelo crab ; been ages since I tried N900 alas... uerm iirc the problem was that for some oid reasons ipv6 packaets weere getting dropped and were not getting the functionality that worked under cssu+power kernel i.e. at least ipv6 autoconfig without userland support, used to work .......
<freemangordon>
Wizzup: but I think I have tested -depth 16 and it is just fine
<freemangordon>
keep in mind this halves the required memory for graphic ops by a factor of 2
<freemangordon>
because 24bpp is actually 32 bpp
<freemangordon>
we have ARGB/XRGB
<freemangordon>
also, increasing CMA > 32 and having compcache will just leave no memory for anything else
<Wizzup>
crab: looks like it's quite small: $ ls -lsh usr/lib/icd2/libicd_network_ipv6.so
<freemangordon>
oh, while we are at it - please, *disable* compcache
<freemangordon>
I've spent a lot of time on that, the final verdict is - unusable
<crab>
Wizzup: fwiw, (and i absolutely don't want to contradict what Pali wrote, because i know hes a "legend in the scene" so to speak),
<crab>
but im not entirely sure that what he said about just having the sysctl is enough is completely true
<crab>
ill see if i can confirm that my settings are sane at some point though
<uvos>
freemangordon: well the fact that i just tried to switch x to 16bit but amdgpu refuses to create 16bit scanout buffers should be an indication how well this is still tested
<bencoh>
:]
<uvos>
freemangordon: from runing 16bit on ddk1.9: anything qml and firefox is realy slow because they do 32 bit rendering regardless and convert to 16 bit for output
<freemangordon>
16bit is broken on ddk1.9
<bencoh>
they probably rely on RGBA for everything GL anyway
<freemangordon>
which is slow on ARM, because it uses BGRA ;)
<freemangordon>
well, at least SGX do
<uvos>
BGRA istent that unusal
<Wizzup>
crab: I took a quick look and libicd-network-ipv6 is only 20kb so RE'ing it probably takes 1-2 days, I don't have that time now but in early 2022 I could take a look
<bencoh>
oh, I didn't mean RGBA vs BGRA, but vs other 16b formats
<bencoh>
like 565
<freemangordon>
uvos: I don;t want to go into lengthy discussions, but I will insist on having 16bpp until someone provides some evidence that 32bpp is better (on mobile)
<Wizzup>
freemangordon: maybe just try it for yourself and see
<freemangordon>
and "app X does not work" is not really an evidence
<freemangordon>
Wizzup: sure, but not now
<uvos>
ignoreing the problem with compatability: i mean better is really subjective anyways
<Wizzup>
if it works better then we can revisit
<crab>
Wizzup: put it on the list
<uvos>
people look at photos and stuff on mobile alot
<crab>
but far down the list would be my suggestion
<freemangordon>
uvos: and how many bits do you think your display has?
<crab>
and if anyone bothers you ill happily tell 'em what i have in my script! ;)
<freemangordon>
bits per pixel that is
<freemangordon>
12?
<Wizzup>
crab: well there are fremantle scripts already
<bencoh>
n900's display is pretty decent actually, but yeah, I doubt it has a super high definition
<Wizzup>
libicd-network-ipv6-scripts
<freemangordon>
n900 is a winner, IIRC it uses 14 or 16 bpp
<freemangordon>
but it is a real GRB
<freemangordon>
*RG
<uvos>
freemangordon: well more that 16 is not usual even on cheep pannels, besides they do temporal dithering
<freemangordon>
aah
<uvos>
obv ainchent suff like n900 is not the benchmark here
<freemangordon>
actually it is
<freemangordon>
because it has a real RGB
* Wizzup
checks out for a bit :P
<Wizzup>
crab: please apt update and apt upgrade
<Wizzup>
crab: wlan should be fixed
<freemangordon>
unlike 'modern' displays which do not
<crab>
roger
<uvos>
freemangordon: ? "real RGB"
<uvos>
you mean not pentile?
<uvos>
who still uses pentile?
<uvos>
except really really high density stuff
<freemangordon>
and what is used now?
<uvos>
rgb
<freemangordon>
provide example, please
<bencoh>
Sony ACX565AKM (n900)
<uvos>
xt1602
<freemangordon>
Last time I checked iphone, they were using pentile
<uvos>
besides pentile is totaly orthagonal to color depth
<bencoh>
Display Colors16.7M (8-bit), CIE193170%
<uvos>
pentile reduces spacial resolution not color resolution
<freemangordon>
uvos: as I told you ,I don;t have time now (not will later :) ) to go into lengthy discussion on that
<crab>
i wonder what was up with the time - zone was correct but it was an hour out. maybe a result of accidentally booting into regular maemo with a weird tz when i pulled the battery before,
<crab>
and then when it got into maemo-leste it was like "meh im not stepping this a whole hour in one big chunk" or something
<uvos>
so qt webbrowser shows a vkb now - but only if the deivce is not opened :)
<Wizzup>
crab: yeah looks fine, maybe try to reboot to make sure it worked
<Wizzup>
or sudo /etc/init.d/icd2 restart
<Wizzup>
uvos: cool, how does that work with qt?
<Wizzup>
the IM is still missing
<uvos>
Wizzup: it uses qt's internal vkb - i allready supported that
elastic_dog has quit [Ping timeout: 268 seconds]
<uvos>
Wizzup: i just added support for checking with mce
<uvos>
to get the slide state
<Wizzup>
hm, him does that normally I think (The slide checking)
<uvos>
Wizzup: yeah ofc
<Wizzup>
as in the application doesn't need to know
<uvos>
yeah but its not him
<Wizzup>
sure
<uvos>
it implements/uses qts own vkb
<Wizzup>
eventually I don't think we want many different vkbs, but having a decent browser is exciting
<uvos>
yeah sure
<uvos>
but rn we dont have any browser that works on ts only devices
<Wizzup>
yeah
<Wizzup>
brb
<crab>
restarting looks good
<Wizzup>
cool
<crab>
ill try a reboot too but since you are brb i will attend to irl job hopefully before you get back!
<Wizzup>
qrc:/QtQuick/VirtualKeyboard/content/components/Keyboard.qml:36:1: module "Qt.labs.folderlistmodel" is not installed
<uvos>
ugh
<uvos>
:P
<Wizzup>
dialogs?
<uvos>
no
<uvos>
it depends on that allredy
<uvos>
qml-module-qt-labs-folderlistmodel
<Wizzup>
you're keeping notes btw, right? :p
<uvos>
yeah im adding it to control as you type
<Wizzup>
ok
<Wizzup>
now it runs
<Wizzup>
it's pretty hard to type on the virtual keyboard hehe
<Wizzup>
hm my touch screen stopped responding
<Wizzup>
works fine in evtest
<uvos>
lock unlock display dosent helP/
<uvos>
?
<uvos>
hildon is responsive otherwise?
<Wizzup>
it did help
<uvos>
hmm
<Wizzup>
but it's stil lweird
<uvos>
yeah
<uvos>
i wonder if we have the latest ts-buttons
<uvos>
i told tmlind to pull
<uvos>
there was a bug that could cause it
<uvos>
ill check later
<Wizzup>
ok
<uvos>
i kicked qtwebbrowser again
<uvos>
i hope its fine now
<uvos>
but really who knows
<Wizzup>
I think we should figure out the dependency things for the droid4
<Wizzup>
5.15 seems pretty ready for -devel
<Wizzup>
what did we say, omap-linux or maemo-linux? I think omap yeah?
<uvos>
i prefer maemo-linux and f**ckl it to devices that cant keep up, but the resonable objection was that we cant expect every device to work with latest linux
<uvos>
so omap-linux
<uvos>
an alternatie would be to make it just linux with different sections
<uvos>
at least when we reorganize the sections to be saner
<Wizzup>
ok
<Wizzup>
so I guess I will add a bootmenu.scr to leste-config for the n900
<Wizzup>
it will be hard to know what the exact fremantle settings are...
<Wizzup>
perhaps the usual attachboot and emmcboot are enough to cover that
<Wizzup>
ideally it would just read bootmenu.scr from all partitions I guess
<Wizzup>
I think what will be the browser is what we invest in
<Wizzup>
for example if we get someone to work on the qml parts of qtwebbrowser to make it more hildon usable, it has a good chance
<uvos>
not sure what you mean
<Wizzup>
but the same is true for surf-gtk
<uvos>
im also not sure QtWebBrowser is ideal
<uvos>
also n900 needs a custom solution i think
<uvos>
also sholes if you ever invest the time to port leste to that
<Wizzup>
sholes?
<uvos>
pre mapphone motorola devices
<Wizzup>
like the xt610?
<uvos>
yes
<uvos>
"for example if we get someone to work on the qml parts of qtwebbrowser to make it more hildon usable, it has a good chance" except vkb im not sure what that would mean
<Wizzup>
the qtwebbrowser isn't in any way hildonised
<Wizzup>
but it seems to be quite fast
<Wizzup>
it needs some ui that makes it look like hildon
<Wizzup>
and a menu where it's title is
<uvos>
oh you want you theme it
<Wizzup>
a fullscreen mode that works well
<Wizzup>
I think it's more than just theming
<Wizzup>
s/it's/its/
<Wizzup>
(for it's title)
<uvos>
not sure what you would want the title menu to contain really
<uvos>
but ok
<Wizzup>
settings
<Wizzup>
bookmarks view
<Wizzup>
useful stuff
<Wizzup>
saved credentials
<Wizzup>
being able to change rotation, full screen, auto focus addr field, memory cache size, how zoom works, password saving, popups, cookies, javascript,
<Wizzup>
that kind of stuff
vifino has joined #maemo-leste
<Wizzup>
sicelo: n900-pm is now in -experimental
<uvos>
i mean the only thing i think it lacks is ad block and plugin support - witch it has if you build it against qt 5.15 or newer
<uvos>
but sure
<Wizzup>
I think some form of customisation is important
<Wizzup>
but I also don't know how well it works since the UI is mostly custom
<Wizzup>
or foreign to me
<uvos>
"how zoom works" zoom works idealy for me
<Wizzup>
I figured out that I had to swipe to see a url bar
<uvos>
other than on n900 where you cant zoom
<Wizzup>
the zoom was weird for me since eventually it just zoomed when I used one finger
<uvos>
but it ooms there at startup anyhow
<Wizzup>
but maybe that was the ts buttons bug
<Wizzup>
fremantle browser does the right thing imho hiding the top bar and having the url at the bottom
<Wizzup>
at least having the bar visible
<Wizzup>
anyway we all have different opinions on this, but I think there's some work to be done to make it a bit more usable
<Wizzup>
and if we have to do that, we might want ot keep that in mind when selecting a default
<uvos>
the point i was making before we should select a default before we start working on something
<uvos>
at least majorly
<uvos>
and a descision on if the n900 just not working is acceptable
<Wizzup>
yes, but then we also need to weigh stuff like gtk2/gtk3/qt5
<Wizzup>
and how much it's maintained by others
<uvos>
yes
<uvos>
yes absolutly
<tmlind>
Wizzup: n900-pm, cool :)
<Wizzup>
tmlind: it works, but there are a few blockers that I don't know how to tackle
<tmlind>
which bits are those?
<tmlind>
spi is used by wlan probably
<Wizzup>
ST_SDRC, ST_OMAPCTRL, ST_I2C1
<Wizzup>
spi only blocks sometimes, so wlan idles
<tmlind>
hmm i2c would block idle for sure
<Wizzup>
yeah I think that's twl?
<Wizzup>
(at least that's what the &i2c1 entry in the dts hints at)
<tmlind>
but that should be active only now and then
<uvos>
twl i2c same thing
<Wizzup>
unfortunately that seems to be most of the phone
<uvos>
i2s is trademarked by philips or something
<tmlind>
usb phy maybe?
<Wizzup>
uvos: I mean the twl subsystem
<uvos>
right ok
<Wizzup>
tmlind: phy_twl4030_usb ?
<tmlind>
yeah try rmmod that
<Wizzup>
I can remove it, but I think musb_hrc depends on it?
<tmlind>
yup
<tmlind>
rmmod omap2430, then the phy
<Wizzup>
hm it reset, me do this properly
<Wizzup>
(just a minute)
<Wizzup>
I had to remove the usb gadget before phy_twl4030_usb would rmmod, but removing the gadget caused a reset
<Wizzup>
3/3 boot failures, we'll get there :p
<Wizzup>
there we go 5th attempt
<Wizzup>
hm, it still thinks it is in use
<Wizzup>
udc_core needs musb_hdrc, so do I remove that too then?