IRC log of #maemo for Sunday, 2017-04-23

*** dafox has joined #maemo00:14
*** florian has quit IRC02:46
*** spiiroin_ has joined #maemo02:50
*** spiiroin has quit IRC02:53
*** hurrian has joined #maemo03:21
*** hurrian_ has quit IRC03:21
*** infobot has quit IRC03:23
*** infobot has joined #maemo03:24
*** LouisA has joined #maemo03:35
LouisAhey guys03:35
*** xy2_ has quit IRC03:48
*** dafox has quit IRC04:03
*** cyphase has quit IRC04:25
*** cyphase has joined #maemo04:30
*** hurrian has quit IRC04:57
*** hurrian has joined #maemo04:58
*** LouisA has quit IRC05:22
*** luke-jr has quit IRC06:19
*** luke-jr has joined #maemo06:27
*** vahe has joined #maemo06:38
*** pagurus` has joined #maemo06:45
*** pagurus has quit IRC06:47
*** pagurus` has quit IRC06:47
*** pagurus has joined #maemo06:47
*** Zungo has quit IRC07:28
*** DocScrutinizer05 has quit IRC07:31
*** DocScrutinizer05 has joined #maemo07:31
*** sunshavi has quit IRC08:11
*** sledges has quit IRC10:34
*** ginggs has quit IRC10:50
*** ginggs has joined #maemo10:51
*** HRH_H_Crab has quit IRC10:52
*** HRH_H_Crab has joined #maemo10:54
*** xorly has joined #maemo10:57
*** Ex-Opesa has quit IRC11:13
*** xorly has quit IRC11:30
*** xorly has joined #maemo11:31
*** xorly has quit IRC11:51
*** xorly has joined #maemo11:52
*** crazyguy` has joined #maemo12:22
*** florian has joined #maemo12:28
*** LouisA has joined #maemo13:12
*** shentey has joined #maemo13:44
*** Pali has joined #maemo13:44
*** shentey has quit IRC13:51
*** shentey has joined #maemo13:51
*** Pali has quit IRC14:16
*** xy2_ has joined #maemo14:42
*** hurrian has quit IRC14:51
*** L29Ah has joined #maemo14:57
*** L29Ah has left #maemo15:13
*** LouisA has quit IRC15:16
*** hurrian has joined #maemo15:21
*** herekun has joined #maemo15:35
*** shentey_ has joined #maemo15:49
*** shentey has quit IRC15:52
*** xy2_ has quit IRC15:56
*** xy2_ has joined #maemo15:57
*** xy2_ has quit IRC16:10
*** xy2_ has joined #maemo16:12
*** dafox has joined #maemo16:18
*** mavhc has quit IRC17:16
*** mavhc has joined #maemo17:19
*** Cor-Ai has quit IRC17:20
*** Cor-Ai has joined #maemo17:24
*** SmilyOrg has joined #maemo18:05
*** Smily has quit IRC18:05
*** auenfx4 has joined #maemo18:16
*** auenf has quit IRC18:16
*** L29Ah has joined #maemo18:18
*** L29Ah has left #maemo18:40
*** L29Ah has joined #maemo18:55
*** shentey_ has quit IRC19:02
*** auenfx4 has quit IRC19:10
*** auenf has joined #maemo19:14
*** xy2_ has quit IRC19:20
*** xy2_ has joined #maemo19:21
*** L29Ah has left #maemo19:25
*** L29Ah has joined #maemo19:25
*** msava has joined #maemo19:47
*** msava has quit IRC19:48
*** msava has joined #maemo19:48
*** msava has quit IRC19:50
*** msava has joined #maemo19:51
*** msava has quit IRC19:52
*** msava has joined #maemo20:03
*** Pali has joined #maemo20:03
*** msava has quit IRC20:03
*** msava has joined #maemo20:04
*** msava has quit IRC20:04
*** mp107 has joined #maemo20:24
*** vahe has quit IRC20:27
*** platicus has quit IRC20:58
*** tanty has quit IRC20:58
freemangordonPali: ping20:59
*** platicus has joined #maemo20:59
*** tanty has joined #maemo20:59
freemangordonanyway, it is not a question only for Pali, so...21:00
freemangordonWizzup: parazyd: while implementing maemo gtk on top of upstream, I am facing a hard to be solved problem how to access various widgets private structure members21:01
freemangordonI can think of 2 options, both of them i don;t really like:21:01
freemangordon1. redefine private structure in my code, call g_type_instance_get_private and call it a day.21:03
freemangordon2. reimplement the whole class instead of just subclassing21:03
freemangordon1. is really, really ugly, but needs less coding21:04
freemangordon2, is not *that* ugly, but I'll have to LD_PRELOAD hildon-gtk and hack in the gtk_whatever_widget functions, which may conflict with GTK3 programs. also, I am not sure how will that work in case of "inter-gtk" calls21:06
freemangordonso, do you have any other idea on how to achieve that?21:07
freemangordonbencoh: KotCzarny: the same question to you, and all other that can help21:07
freemangordon*all others21:07
KotCzarnyarent private members meant to be private?21:07
freemangordonyes, but I am trying to act like a part of gtk itself21:08
KotCzarnyie. either find interface/func to access them or it might fall apart/be incompatible21:08
*** tanty has quit IRC21:08
*** platicus has quit IRC21:08
freemangordonyes, I know21:08
freemangordonthus "really, really ugly"21:08
KotCzarnygtk has a concept of custom widgets?21:08
freemangordonI am trying to implement hildon gtk, without patching gtk code :)21:09
*** platicus has joined #maemo21:09
*** tanty has joined #maemo21:09
KotCzarnyand you want to overload them in existing apps or just add hildon functionality?21:09
freemangordonboth21:10
KotCzarnyhrm21:10
freemangordonas adding hildon functionality requires accessing gtk internals21:10
freemangordonKotCzarny: see https://github.com/community-ssu/gtk/blob/master/gtk/gtkiconview.c#L156521:11
freemangordonfor example21:11
KotCzarnyhow about writing gtk-hildon port and provide it as an alternative (while being a requirement for hildon enabled apps/)21:11
bencohfreemangordon: how does it work thus far then?21:11
freemangordonbencoh: where?21:11
bencohand by "upstream" do you mean gtk2 or gtk3 (or doesn't matter)?21:11
freemangordongtk221:11
KotCzarnyin a way jpegturbo 'overloads' libjpeg21:11
bencohfreemangordon: on current maemo/gtk221:11
freemangordonbencoh: code is heavily patched21:12
KotCzarnybencoh: on maemo nokia just forked gtk21:12
freemangordonKotCzarny: I don't want to fork gtk21:12
bencohfreemangordon: gtk on maemo is heavily patched?21:12
KotCzarnybencoh: yup21:12
freemangordonKotCzarny: re "in a way jpegturbo 'overloads' libjpeg"21:13
bencohI see :/21:13
freemangordonbencoh: just look at GtkIconView code ^^^21:13
KotCzarnyfmg, why not? gtk2 wouldnt receive updates afair21:13
bencohfreemangordon: aww.... that's not pretty :/21:14
KotCzarnyand it would allow you to do things more elegant21:14
freemangordonKotCzarny: because, this is lots of work, I'd rather avoid if possible21:14
KotCzarnyfmg, but you are doing it anyway, no?21:14
freemangordonKotCzarny: no21:14
KotCzarnyi dont understand why do you need private members then and the 'implement hildon functionality'21:15
*** mp107 has quit IRC21:15
freemangordonor rather - I am doing maemo part only, and if I make it that way, it won;t depend on gtk2 version21:15
freemangordonKotCzarny: look at the code ^^^21:15
KotCzarnycant at the moment, console only21:15
KotCzarnywithout mouse21:15
bencohah :]21:16
bencohcopypaste in screen! :D21:16
KotCzarny:)21:16
*** florian has quit IRC21:16
freemangordonwill explain - for example GtkIconView has a list of "items" which are in priv struc only, without accessor functions21:16
freemangordonhildon gtk draws a tick mark on every selected item21:17
KotCzarnyhmm, afair you can get access to items in tree/list store21:17
freemangordonKotCzarny: how?21:17
freemangordonyou get some cryptic GtkTreePath or somesuch21:17
KotCzarnyby accessing it's tree/list store?21:17
KotCzarnyyeah, its hell to understand, but you can21:18
freemangordonthere is no way to access it, at least I was unable to find one for the last 3 days21:18
KotCzarnycheck oscp-remote-gtk21:18
KotCzarnyi have some helper funcs there21:18
KotCzarnynot the best code in the world but..21:18
KotCzarnylet me check the link for you21:18
freemangordonwhere is the code, on github?21:19
Enrico_MenottiHello. I have Devuan booting on the n900. The system image is one I just debootstrapped on Devuan rc which I installed yesterday on my old laptop. But I have problems with the kernel. Someone here (I don't remember who exactly was, if freemangordon or parazyd ) had given me another kernel image. If I use that one, Devuan boots. However, with a kernel I built a few days ago, and also with one built right now, Devuan21:19
Enrico_Menotti does not boot (but Debian does). The kernel boots, but does not find selinux; it stays there for a while and then the n900 reboots. If I give init=/bin/sh I get a shell but it's unstable - after a few seconds the n900 reboots.21:19
KotCzarny31.135.195.151:20280/arch/oscp/21:19
KotCzarnysorry, typed that manually21:19
KotCzarnyit's a source package prepared for arch21:20
freemangordonKotCzarny: thanks, going to dig in it21:20
KotCzarnyoscp-remote-gtk is in libs/_tools/21:20
KotCzarnyit's hackish, but works for me21:20
freemangordonEnrico_Menotti: sorry, don;t have time now21:20
bencohis wdog enabled at boot?21:21
Enrico_Menottifreemangordon No problem.21:21
KotCzarnyaccessing items in both, via iterators and via element reference21:21
bencoh(that'd be strange, but ... who knows)21:21
Enrico_Menottibencoh Is there a parameter to enable the watchdog manually?21:22
freemangordonKotCzarny: any hint where to look at? as it is lots of code there21:22
KotCzarnyyou probably want to look at oscp-remote-gtk.c func tv_show()21:22
KotCzarnyand tv_click()21:23
*** eMHa_ has quit IRC21:24
freemangordonKotCzarny: there is no access to the underlying items, just to the path21:25
freemangordonand I didn;t find a way to get dimensions and position of cells in GtkIconView21:25
freemangordonand I need  those to be able to draw a tick GdkPixbuf21:26
KotCzarnywhy not defining your own widget then?21:28
freemangordonKotCzarny: because all maemo applications will have to be ported to use it :)21:28
KotCzarnysee? fork it21:28
KotCzarny:)21:29
freemangordonhehe21:29
freemangordonI'd rather avoid that if possible21:29
*** eMHa_ has joined #maemo21:30
*** florian has joined #maemo21:30
KotCzarnymaybe just replace the icon with tick drawn over it?21:31
freemangordonKotCzarny: I cannot get the coordinates of the bounding box of the cell to draw into21:32
freemangordonotherwise there is no problem to draw one more icon21:32
KotCzarnynot one more icon, replace the image21:32
freemangordonI cannot do that, see http://talk.maemo.org/showpost.php?p=1527039&postcount=16821:33
KotCzarnywill check links tomorrow21:33
freemangordon"Activate views" is GtkIconView21:34
bencohEnrico_Menotti: no idea actually, that was just an idea21:36
Enrico_Menottibencoh Actually I think that be the problem. But I don't know why.21:36
parazydEnrico_Menotti: you have to load the watchdog modules on boot, or include them in the kernel21:39
bencohah :)21:39
parazydi don't remember the names, but grep the kernel config for WATCHDOG21:39
Enrico_Menottiparazyd So your kernel has the modules included?21:40
Enrico_MenottiAlso, why with Debian no problem?21:40
parazydyes, the config is here: https://github.com/dyne/arm-sdk/raw/next/boards/kernel-configs/n900.config21:41
parazydi don't know, i never tried debian21:42
* DocScrutinizer05 wonders if ANYBODY _ever_ will tackle the relatively (in relation to the above) simple task of just building a maemo5 armel image based on debian/devuan from scratch21:42
DocScrutinizer05without massive porting to new kernels, new libs, new arch21:42
parazydfreemangordon: you cannot get the coordinates of the tick box or the parent?21:44
freemangordonparazyd: of the tick box21:47
freemangordonparazyd: or to make it more clear - I have to draw tick mark in every selected cell21:48
parazydah, so you don't know the cell sizes21:48
freemangordonmhm21:48
freemangordonand positions21:48
parazydno, i thought just spawning the tick box on the opposite of 0,021:49
parazydnevermind then :D21:49
freemangordonmaybe I can calculate those, but ...21:49
freemangordonhmm, maybe this is the correct way to do it21:49
DocScrutinizer05I'm afraid this whole porting thing leads nowhere near anything that would be capable of running apps from maemo-extras (unless somebody would recompile them, which obviously is a mad idea to do)21:52
freemangordonDocScrutinizer05: actually those should run without recompiling, at least on x8621:53
freemangordonand on armel as well21:54
KotCzarnyfork the lib, told'ya21:54
parazydaren't the libs they are linked to much older than current?21:55
KotCzarnyfor gtk2 you are quite safe21:55
freemangordonparazyd: so?21:55
freemangordonKotCzarny: and what about gtk3?21:55
parazydfreemangordon: maybe stuff is gone21:55
KotCzarnyfor gtk3 you have to rewrite apps anyway21:55
KotCzarnynot even recompile would help21:55
freemangordonif we talk hildon stuff, what is there is decided by us21:56
freemangordonbut yeah, there might be removeed function21:56
freemangordon*removed functions21:56
parazydyeah i think DocScrutinizer05 is talking about _all_ the software from the repos... anyway. /end bikeshedding21:57
freemangordonparazyd: what about having forked gtk2 in devuan-maemo repo? what is the policy?21:57
parazydfreemangordon: you're free to do so as long as you keep backwards compat21:57
parazyd(because of everything else using gtk2)21:57
freemangordondefine backwards please21:57
KotCzarnyplain gtk2 apps21:57
freemangordonas it will change the way gtk apps look21:57
freemangordondo we talk ABI only here?21:58
parazydabi21:58
parazydyes21:58
freemangordonhmm21:58
parazydthe look isn't the problem, i mean everything will be running inside hildon anyway21:59
parazydit's just the calls21:59
KotCzarnyas a bonus, those old gtk2 apps might get few new features, fe. pannable areas? ;)21:59
bencohwaitamin, there is an independent devuan-maemo repo?21:59
KotCzarnypretty handy with touch screens22:00
freemangordonbencoh: WIP22:00
parazydbencoh: there will be22:00
DocScrutinizer05that's the idea22:00
parazydfreemangordon: we'll be free to experiment in any case. that's the point of the thing. the repo is separate from devuan, and we just include the pacakges we want22:01
parazydthen they take precedence over devuan's22:01
DocScrutinizer05just like CSSU22:01
DocScrutinizer05well, technically different, but same effect22:02
Wizzupfreemangordon: I'd be fine with (1) for now22:02
freemangordonparazyd: well, forward-porting all hildon gtk stuff for the sake of experiment... a bit of an overkill :)22:02
WizzupI'd go with relatively little effort for now, unless it means -a lot more work- in the future22:03
Wizzupthis just sounds like an ugly hack, but not something that will bite us anytime soon22:03
freemangordonparazyd: ah, BTW which of those 2 options you dislike less?22:03
DocScrutinizer05Wizzup: that's a good approach22:03
freemangordonWizzup: yes, gtk2 private structures won;t change22:03
parazydalways less code22:03
freemangordonthough I am not sure it will be less code, as I'll have to override almost all of the class functions22:04
DocScrutinizer05((<parazyd> yeah i think DocScrutinizer05 is talking about _all_ the software from the repos...)) of course. And I don't see freemangordon not doing so22:05
DocScrutinizer05maybe I missed the point22:05
*** LouisA has joined #maemo22:05
DocScrutinizer05that's why I insist in armel instead armhf images22:05
bencoh...22:06
DocScrutinizer05which are admittely a tad less effective, but according to studies quoted on debian wiki, "up to 40% faster for **heavily numbercrunching** apps" OWTTE22:07
freemangordonWizzup: parazyd: the correct way is to fork gtk2, this is for sure. Maybe I'll just bite the bullet and do it.22:07
freemangordonparazyd: If I choose that route, do you know which git repo shall I use as a base? debian? or there is devuan repo for all the stuff?22:09
*** florian has quit IRC22:09
parazydwe haven't forked gtk222:09
parazydbut you should do an apt-get source from devuan to be sure you have the right thing22:10
DocScrutinizer05while possibly it's irrelevant for the particular app if itself is softfp or hf build, and different apps may coexist, I see problems in libs and in task scheduler when both are used in parallel22:10
*** xes_ has joined #maemo22:12
DocScrutinizer05if we could maje sure all lib ABIs are agnostic of softfp vs hf and the task switcher in scheduler will take care of *all* registers (incl FP), it *might* even work to run armel apps in a hf environment, as long as the softfp libs are available22:12
DocScrutinizer05disclaimer: AIUI22:13
DocScrutinizer05(if softfp are even shared object libs and not embedded code by gcc during build time)22:14
freemangordonWizzup: parazyd: seems I am not the only one https://git.javispedro.com/cgit/topmenu-gtk.git/tree/module/menuitem-proxy.c#n138 :)22:14
freemangordonand this is in debian repos22:14
*** xes has quit IRC22:15
freemangordonok, going the hacky way for now22:15
DocScrutinizer05freemangordon: would you know if the armel scheduler saves (and restores) the FP registers on task switching?22:16
DocScrutinizer05also I hope all kernel tasks preserve FP as well22:17
freemangordondon't know, sorry, but in theory all regs should be saved22:18
DocScrutinizer05ack, thought as much22:18
freemangordonalong with all the state22:18
DocScrutinizer05trying to wrap my head around that damn soft(fp) vs hf thing22:19
DocScrutinizer05and armel vs armhf22:19
KotCzarnyfmg: why not maemo's gtk?22:19
DocScrutinizer05I recall back with meego it was like 2 weeks of arguing22:19
freemangordonKotCzarny: in what regard?22:20
freemangordonwhat to do with maemo's gtk?22:21
*** florian has joined #maemo22:22
KotCzarnyto take it as a fork base22:25
freemangordonit is old22:25
KotCzarnyit is, unfortunatelly22:26
DocScrutinizer05it worked, so far22:26
KotCzarnyquite a lot of packages would require newer one22:26
DocScrutinizer05there's always "quite a lot of packages that require newer than X version of Y" out there22:27
KotCzarnyas in: most of them (from devuan binary repos)22:27
*** mp107 has joined #maemo22:50
*** mp107 has quit IRC23:13
*** sunshavi has joined #maemo23:26
*** shentey has joined #maemo23:43
*** dafox has quit IRC23:52

Generated by irclog2html.py 2.15.1 by Marius Gedminas - find it at mg.pov.lt!