**** BEGIN LOGGING AT Tue Jan 24 02:59:57 2012 Jan 24 03:07:22 hmm Jan 24 03:07:30 save/restore cursor is not having correctly Jan 24 03:17:37 dtzWill, i know see why i need a struct Jan 24 03:17:38 or Jan 24 03:17:40 should use one ;) Jan 24 03:42:15 dtzWill, ping Jan 24 03:45:03 pong-ish Jan 24 03:45:06 what can i do ya for? Jan 24 03:49:21 dtzWill, pushed my alt screen stuff if you wanted to review Jan 24 03:49:26 fairly simple Jan 24 03:49:39 i just dont if what I did could cause memory issues Jan 24 03:49:53 PuffTheMagic: won't be able to do a good review until some time tomorrow, sorry Jan 24 03:50:00 k Jan 24 03:50:28 (that said I might get too curious for my own good before then, we'll see :P :D) Jan 24 03:52:56 dtzWill, its a small patch, if u can green light it tonight I can tag 3.0.4 Jan 24 03:53:03 want to get this beast out to the pub Jan 24 03:53:08 its got shit tons of improvements Jan 24 04:03:24 PuffTheMagic: offhand seems reasonable to me. no need to explicitly call copy constructor and then use an assignment operator; "m_savedScreen.m_data = m_data;" should do the neccessary magic Jan 24 04:03:54 memory-wise it's fine, but i'm concerned about things like clearing it during reset(), and invalidating it when changing screen resolution or what-have-you Jan 24 04:05:39 j/w what does this alt-screen get us? Jan 24 04:05:45 (what application(s) use it?) Jan 24 04:07:41 htop uses it for one Jan 24 04:07:48 lets say you have shit on your screen Jan 24 04:07:53 like you ran some commands Jan 24 04:07:59 then launched htop Jan 24 04:08:01 when u exit Jan 24 04:08:07 u get your old shell contents back Jan 24 04:08:44 clearing what during reset? Jan 24 04:09:05 vim uses it Jan 24 04:09:07 oh idk should we clear the saved screen when doing a reset? Jan 24 04:09:24 i see, and yeah that's useful for quick control-z/fg'ing of a vim process :) Jan 24 04:09:31 didn't realize that was alt-screen being used, nifty Jan 24 04:09:41 we "could" clear it Jan 24 04:09:50 idk what the proper behavior Jan 24 04:09:56 idk, i'm just saying idk about those kinds of cases Jan 24 04:09:58 this is again one of those illdefined xterm features Jan 24 04:10:04 * dtzWill nods Jan 24 04:10:11 it passes the vttest's Jan 24 04:10:15 so im happy with that for now Jan 24 04:10:24 this will not cause crashes, afaik, with the *possible* exception of if you expand the height of your terminal Jan 24 04:10:33 and then swap out a saved screen Jan 24 04:10:48 should I save the screensize variable too? Jan 24 04:10:52 the saved screen might not have enough lines, and while we deal gracefully with a line being too short, we die if there aren't enough lines Jan 24 04:11:05 that, or just add a check to ensure it's at least the proper height Jan 24 04:11:16 and either add padding or bial/discard the backup if it's not Jan 24 04:11:29 but i mean in terms of shipping it that's not a show-stopper Jan 24 04:11:40 at least not imo, not at this active-dev early stage Jan 24 04:11:55 0.3.4: Security fixes (defaults to non-root user), key repeat, Dvorak and German vkb layouts, just type action, multiple instances/cards, custom exec commands, auto relaunch, new escape code parser Jan 24 04:12:00 am i missing anything Jan 24 04:12:39 does github show tags inline on the commit history? Jan 24 04:12:56 no :/ Jan 24 04:12:58 i wish it did Jan 24 04:13:00 dtzWill, Jan 24 04:13:09 u can do git log v0.3.3..HEAD Jan 24 04:13:16 yeah i just had github open already Jan 24 04:13:18 haha Jan 24 04:13:19 :3 Jan 24 04:14:01 meh, rendering improvements (both faster (see cmatrix running smoothly) and glitch fixes (see fullscreen)) but not sure that's end-user worthy Jan 24 04:14:08 (since there have been no bug reports on those issues) Jan 24 04:14:13 sounds good to me :) Jan 24 04:14:49 yeah, those can be safely thresholded into the git commit log ^.^ Jan 24 04:14:54 added the render improvements Jan 24 04:15:33 and tagged Jan 24 04:15:44 haha \o/ Jan 24 04:18:19 dtzWill, did u notice i tweaked the about popup a tad Jan 24 04:18:41 i did not! Jan 24 04:18:42 * dtzWill checks Jan 24 04:18:57 nothing big Jan 24 04:19:01 still needs more work Jan 24 04:19:23 (it now has links for project/donate/issues?) Jan 24 04:19:29 ya Jan 24 04:19:33 instead of debug info Jan 24 04:19:34 :D great. Jan 24 04:19:45 yeah, that's useful and a good place to have that stuff Jan 24 04:20:01 if theres stuff you want there specifically just say so Jan 24 04:20:06 or add it your self Jan 24 04:20:21 okay, sounds good and thank you :). Jan 24 04:20:42 bah, the keys /are/ noticeably less responsive than earlier, no? Jan 24 04:20:44 i mainly used the about popup to get the rows/cols before i udpated vttest to figure them out automatically Jan 24 04:21:04 haha yeah, which is of course fine. no one minds if there's also misc debug information like git version or something Jan 24 04:21:25 in fact, for my own sanity, i mean add that. it would mean i'd have to actually test using the ipk methods :P :D Jan 24 04:21:31 *i might Jan 24 04:22:03 err later. i've seen projects do nifty things like have the 'proper' version if it's tagged, otherwise list it as devel and give a git commit id Jan 24 04:22:04 ummm input might be a tad slower Jan 24 04:22:08 i cant tell exactly Jan 24 04:22:43 dtzWill, ya i did something like that for Tide but it gets annoying requiring git Jan 24 04:23:05 i usually have such fancy things Jan 24 04:23:17 errr you mean annoying to require git to install a git-hosted project from source? Jan 24 04:24:18 playful harshnesss aside, i do mean the question seriously :) Jan 24 04:24:23 dtzWill, ya, i did something like that fancy in 2 python projects and it was a pain when git or make was not available ;) Jan 24 04:24:44 but typically im the only one who likes those things Jan 24 04:24:58 oil is certainly not turned on by git hashes in version numbers :D Jan 24 04:25:22 well more than once i've been like $#$#%! am i really running what i think i am Jan 24 04:25:32 an since i have two versions installed that gets moreso complicated Jan 24 04:25:34 haha Jan 24 04:25:40 i can add something in there Jan 24 04:25:56 its easier in C and than for python Jan 24 04:27:20 -DVERSION=$(git describe --tags --always) Jan 24 04:34:23 I officially hate opengl Jan 24 04:41:26 windows really hates src/vttest/makefile and I have no idea why Jan 24 04:41:28 * Brybry shrugs Jan 24 04:42:58 us.ryanhope.wterm_0.3.4-1_arm.ipk 24-Jan-2012 04:17 2240650 Jan 24 04:43:01 wtf Jan 24 04:43:05 dtzWill: I want to do something like https://github.com/Brybry/wTerm/commit/76c18e58a5b6691f54277a7070f7fa168ea273b7#L2R371 Jan 24 04:43:13 us.ryanhope.wterm_0.3.3-1_arm.ipk 20-Jan-2012 01:33 931086 Jan 24 04:43:17 debug symbols? Jan 24 04:43:42 shouldnt be Jan 24 04:44:19 did you add more screenshots? Jan 24 04:44:39 screenshots should be not included Jan 24 04:44:42 ohhhhhh Jan 24 04:44:52 autobuilder doesnt use excludes.txt Jan 24 04:45:09 thats probably it Jan 24 04:50:53 anyway, that doesn't actually work/the output gets garbled and thus I assume swapbuffer does more than just copy the back buffer to the front Jan 24 04:51:36 Brybry: i'll be happpy to look at it when i get some spare cycles Jan 24 04:51:55 thanks, it's pretty short Jan 24 05:02:05 after reading up a lot I'm at the point where I think I'm supposed to create a frame buffer object (via opengl es 1 extension) and then draw just changes to that and then draw the FBO texture to the display buffer Jan 24 05:02:39 but I dunno if that's correct at all Jan 24 05:02:40 * Brybry shrugs Jan 24 05:14:20 oh damn Jan 24 05:14:34 people are seeing the 'undefined enyo' error i was seeing before Jan 24 05:14:38 sorry puff and now you're gone :( Jan 24 05:15:11 the path was changed in some commit, i just changed it back locally. i just assumed that was fine since i was scp'ing stuff over and figured something must patch it up since he uses the ipk install method Jan 24 05:15:20 (curious how that WASN't caught, actually) Jan 24 05:15:27 but d'oh Jan 24 05:17:35 no Jan 24 05:17:37 im here Jan 24 05:17:39 i will fix it Jan 24 05:18:04 haha kk hi :) Jan 24 05:18:52 git blame says its stbuehle1 's fault :D Jan 24 05:19:01 haha Jan 24 05:19:07 are you guys using the alpha and beta feeds? Jan 24 05:19:33 (I don't know) Jan 24 05:19:41 is this out of beta? haha Jan 24 05:20:31 well, it seems pushing something to the public feeds where the packaged item hasn't even been tested for being able to run is a bit silly Jan 24 05:20:42 I mean that's why we have alpha and beta feeds, right? Jan 24 05:21:10 it has been tested Jan 24 05:21:17 idk why i dont get this bug Jan 24 05:21:25 dtzWill never mentioned it ot me Jan 24 05:21:35 by tested, I mean tested by someone *other* than the person who packaged it Jan 24 05:21:57 PuffTheMagic: sorry :(. i meant to... :3 Jan 24 05:22:05 I mean "git mv apps/foo beta-apps/foo" is not really that hard ... Jan 24 05:23:03 im sure the 4 wterm users wont mind Jan 24 05:23:39 correct, but you'll piss off 100% of your userbase with just 4 people getting a bad install Jan 24 05:24:07 and give WebOS Internals a reputation for broken public releases in the process Jan 24 09:26:43 PuffTheMagic: well, palm-package fixes the path (i didn't intend to commit that, just happened - but it was useful to debug the vkb with chromium) Jan 24 11:43:01 PuffTheMagic: you know that *_t is reserved by posix? Jan 24 15:01:03 stbuehler, no i didnt realize it was reserved Jan 24 15:11:43 stbuehler: what I do is keep around an index_test.html that's untracked :) Jan 24 15:28:44 so u are running this in Chrome? Jan 24 15:29:00 holy shit wterm is using more cpu Jan 24 15:29:33 htop uses 10-15% more cpu now Jan 24 15:39:01 working on unicode right now Jan 24 15:39:28 i fear my utf8 part in seqparser isn't working yet Jan 24 15:56:16 ok, utf-8 output works. next question: why doesn't work utf-8 input :) Jan 24 16:02:21 sh reads utf-8, and prints '?' instead Jan 24 16:03:54 8bit isnt working yet Jan 24 16:04:31 ahem Jan 24 16:05:02 i can do echo "ä" > /dev/pts/1 (which is the pty slave for my current wterm session) and get the "ä" Jan 24 16:05:47 and wterm also writes the correct utf8 codes Jan 24 16:05:56 only sh doesn't work, and i have no idea why Jan 24 16:06:12 environment vars look fine (same as in novaterm, where utf8 is working) Jan 24 16:06:43 stbuehler, did you create a special glyph cache for all the Umlauts? Jan 24 16:07:49 i made the unicode ranges configurable Jan 24 16:07:53 in 128-char slots Jan 24 16:07:55 i was just emailed an AZERTY french vkb layout Jan 24 16:49:09 stbuehler: do other terminals work with bare sh? maybe busybox's sh doesn't support it or something? Jan 24 16:50:13 stbuehler: also some of the typedef'ing to *_t is my work Jan 24 16:50:24 did not realize that was reserved by anyone, not sure where i picked it up. Jan 24 16:50:45 and yes very clearly reserved, heh. Jan 24 16:53:03 dtzWill, everyone does it Jan 24 16:53:07 i see _t used all the time Jan 24 16:53:26 i think people think its good style :D Jan 24 16:53:31 yeah, and posix can reserve them all they want doesn't mean our code breaks :) Jan 24 16:53:43 but still, good to know such things :) Jan 24 16:53:55 not like posix is gonna define its own TSCell_t or TSScreen_t ;) Jan 24 17:38:20 well, i just wanted to mention _t so you know what happend if my commits change them :) Jan 24 17:38:54 sh works in novaterm (which is a terminal afaik), iirc ssh worked too Jan 24 18:24:08 don't suppose i could request someone hook up the bt 'keyboard dismiss' button to toggle wterm's vkb? Jan 24 18:24:45 i imagine it's straight-forward (at the PDK C level they have a special keysyn for it, PDK_KEYBOARD_DISMISS or something) Jan 24 18:24:56 but i don't know anything about all that enyo stuff :3 Jan 24 18:30:14 dtzWill, I don't think it actually fires PDK_KEYBOARD_DISMISS Jan 24 18:30:25 that's for like if you hit the palm VKB dismiss button Jan 24 18:32:11 I could be misremembering though Jan 24 18:32:54 oh i think you're right, darn. Jan 24 18:33:12 is the bt keyboard dismiss just 'escape' then? guess i can't ask anyone to patch that to 'toggle vkb'... Jan 24 18:33:29 but just that esp with the changes in 3.0.5 to facilitate effective webos navigation via keyboard.... Jan 24 18:33:47 just type->wterm->yay, except now i have half the screen with a vkb on it and have to dismiss it manually Jan 24 18:33:52 well if the native vkb is disabled Jan 24 18:33:58 and in manual mode Jan 24 18:34:04 we might be able to to catch that event Jan 24 18:34:29 I have the code to autodismiss on bt keyboard detection but it requires linking lunaservice, glib, mjson, goodfork, and gthread (though the code itself is rather short) Jan 24 18:34:45 and so I didn't think it was worth that one feature Jan 24 18:34:55 maybe. i seem to recall that event is omnomnom'd by luna anyway, and it won't so much as emit an escape keyevent to SDL apps Jan 24 18:35:15 heh, maybe not :). well just something to think about, although sounds like you've explored that direction a good deal already Jan 24 18:36:20 Brybry, dont forget Ben said they have new bt apis for HID/keyboard stuff post 305 Jan 24 18:36:28 did you ever get a chance to look at that commit from last night? Jan 24 18:36:31 they don't even need new bt apis ._. Jan 24 18:36:38 sure they do Jan 24 18:36:43 they just need to make that one BT notification bus thing private instead of public Jan 24 18:36:48 err public instead of private* Jan 24 18:37:14 ok, not new api Jan 24 18:37:21 but there are going to be changes to the bt api Jan 24 18:37:23 hows that Jan 24 18:37:26 I'd rather they fixed ctrl+a :( Jan 24 18:37:54 Brybry, you have to ask them, they arent psychic :D Jan 24 18:37:56 and ctrl+c and ctrl+v and ctrl+h for standalone PDK apps Jan 24 18:38:38 standalone or not, there should be a function to call that tells webos to handles those or not Jan 24 18:38:43 https://gist.github.com/1646976 Jan 24 18:39:19 Brybry, why do you test in standalone mode so much? what does that offer you? Jan 24 18:40:00 nothing really, I'm just a completionist Jan 24 18:41:07 and it's just good knowledge to know the different behavior between plugin/non-plugin Jan 24 18:42:58 heh, standalone would fix my vkb suggestion :) Jan 24 18:43:13 but lose some features, course Jan 24 18:43:38 we could also just add a fn+whatever key combination to hide it Jan 24 18:44:10 yeah suppose so. 'cept what's 'fn' on the hp bt keyboard? :3 Jan 24 18:44:16 oh, yeah Jan 24 18:44:59 could use ctrl+alt combo though Jan 24 18:46:32 and make wterm save/restore vkb status between sessions Jan 24 18:47:43 Brybry, make a issue for that and I will do it after im done with this Jan 24 18:48:01 it's like 3 lines of code :P Jan 24 18:48:03 but sure Jan 24 18:48:12 just a reminder Jan 24 18:49:58 tyty Jan 24 18:50:06 im converting a paper back into latex Jan 24 18:50:08 :( Jan 24 18:50:27 wrote my masters thesis in latex, my advisor started reshaping it to submit to a journal Jan 24 18:50:34 so copied it all into word Jan 24 18:50:38 and now i have to go back to latex Jan 24 18:50:41 so annoying Jan 24 18:52:28 lol! booo word Jan 24 18:52:47 latex made me not hate writing things. suddenly writing a paper/report/presentation is just like /programming/ your report ;) :D Jan 24 18:56:29 i dont like when people relate latex to programming Jan 24 18:56:40 you need to get pretty advanced till its anything like that Jan 24 18:57:14 its more like wordperfect for dos :D Jan 24 18:59:11 lol Jan 24 18:59:26 fair enough. all i know is that i like the precision of it, and like that i /compile/ my document Jan 24 19:00:24 i like not having my document get fucked up when i cut and paste Jan 24 19:55:59 stbuehler, hows that utf8 stuff going? Jan 24 20:32:36 PuffTheMagic: i'm back now, going to eat and then back to work (on utf8) Jan 24 20:47:34 re latex: i like having my documents in git with readable diffs :) Jan 24 20:48:24 i do that too :D but it also requires hard line breaks Jan 24 20:48:31 and not all editors do that Jan 24 20:56:19 (very much agreed re:throwing them in git and useful diffs) Jan 24 21:00:54 lol Jan 24 21:00:57 i'm so stupid :) Jan 24 21:01:02 utf8 writing was broken Jan 24 21:04:27 WebAppMgr is killing my cpu now (wtf.) Jan 24 21:04:48 wterm 50%, cmatrix 20%, WebAppMgr 90% Jan 24 21:17:33 ipkg-opt doesn't install dependencies? Jan 24 21:18:37 it should Jan 24 21:18:48 puffTheMagic: how much cpu does wterm use when running cmatrix for you (roughly)? Jan 24 21:18:55 trying to figure out if I'm actually getting gains here Jan 24 21:19:18 Brybry, it seems to depend how many CPUs are active Jan 24 21:23:16 stbuehler, nice work Jan 24 21:28:07 with what I'm doing I go from like 75% cpu in top while running cmatrix to 35% but I'm messing something up when I'm drawing the FBO texture to the screen Jan 24 21:28:11 it's all faded :( Jan 24 21:29:29 so for those people hacking on the rendering stuff, here is something that would be nice Jan 24 21:29:35 center the output Jan 24 21:29:50 when it calculates the max rows,cols there is always extra space on right and bottom Jan 24 21:30:03 might look better to center the rendering Jan 24 21:39:05 I just forgot to call glColor4f(1,1,1,1) :D Jan 24 21:39:15 so my alpha wasn't set I guess Jan 24 21:39:16 :D Jan 24 21:41:28 so u got some serious perf improvements? Jan 24 21:43:04 needs a lot of cleanup still but I think so Jan 24 21:49:12 basically I make a frame buffer object that's attached to a texture and then draw only the cells that change every redraw() and then draw that texture to the backbuffer/for swap buffers Jan 24 21:49:25 instead of redrawing every single cell every time Jan 24 21:50:47 I mean, technically it still is refreshing the whole screen but it's not having to call printCharacter() which is all I care about Jan 24 21:54:23 Brybry: i think my next step will be trying the shader thing Jan 24 21:54:27 that should really speed things up Jan 24 21:54:53 well Jan 24 21:55:11 you only update the "texture" that contains character data (and color info), and draw it again Jan 24 21:55:33 but i guess your work on getting the correct events on which chars need to be redrawn will be very helpful for that Jan 24 21:56:15 the "something might have changed - let's redraw everything" should go away :) Jan 24 21:56:18 isn't opengl es 1 compatibility just a wrapper around opengl es 2? Jan 24 21:57:08 someone else said that :) Jan 24 21:58:49 hehe Jan 24 21:58:50 merging this into the latest code is probably not going to be fun Jan 24 21:58:58 :P Jan 24 21:59:22 äöü umlaut test Jan 24 21:59:46 hm, nope. doesn't work -.- Jan 24 21:59:48 I think it might be easier to just make a new branch and then copy/paste stuff instead of dealing with the merge conflicts Jan 24 22:00:12 you should rebase your stuff Jan 24 22:09:13 there's still tons of stuff that could be improved for performance, like instead of flagging cells 'dirty' and then every redraw going through every cell and seeing which were dirty we could push dirty cells to a queue/list Jan 24 22:09:50 a flag in the cell state would probably be enough Jan 24 22:09:54 and for things like erasing we could probably group up cells into big rects or something when it makes sense (and I think a lot of the erasing code is super inefficient) Jan 24 22:10:05 but right now flagging the cell was easiest Jan 24 22:10:08 like 3 lines of code Jan 24 22:10:30 i suspect you may have missed some places :P Jan 24 22:10:57 but flagging cells should be enough. or a combination Jan 24 22:11:12 queue up to 20 cells, then drop the queue and just iterate over all of them Jan 24 22:11:13 I only saw two places where cells had values set Jan 24 22:11:22 sometimes cell values get moved Jan 24 22:11:29 either new rows, or inserting a col Jan 24 22:11:35 when empty cells are created and when a cell value is set Jan 24 22:29:44 cursor definitely isn't redrawing properly though :( Jan 24 22:30:32 up/down/left don't reset the old cells but moving right does Jan 24 22:30:34 which kind of confuses me Jan 24 22:33:14 damn, you're right Jan 24 22:33:17 I missed things! Jan 24 22:33:22 * Brybry sets to whacking Jan 24 22:35:08 stbuehler, got a sec, have a font question Jan 24 22:49:31 PuffTheMagic: sure Jan 24 22:50:50 stbuehler, so to make cmatrix really badass we need to suppor loading fonts Jan 24 22:51:02 so if I have a bitmapped pcf font Jan 24 22:51:17 what would be the best way to get it into wterm Jan 24 22:51:37 i guess font loading is done via esc sequences too? Jan 24 22:51:48 no Jan 24 22:52:16 switching between them is Jan 24 22:52:30 loading font is done by the setfont app Jan 24 22:52:40 or by consolechars app Jan 24 22:52:44 uhm Jan 24 22:52:46 setfont is in the kbd package Jan 24 22:52:53 it uses an ioctl or something Jan 24 22:52:57 and those apps communicate with the terminal over which interface? Jan 24 22:52:58 ah Jan 24 22:52:59 hm Jan 24 22:53:11 thats for real consoles" Jan 24 22:53:25 xterm has another way i believe Jan 24 22:53:32 well, first i want to get shaders running :D Jan 24 22:54:26 stbuehler, you gonna convert to GLES2? Jan 24 22:55:13 i'll try :) Jan 24 22:56:08 if you do, i need to create a new stubby lib for GLES2 so that the autobuilder build it Jan 24 22:56:51 well, first let me see whether i get it working Jan 24 23:00:46 PuffTheMagic, sdlterminal.cpp line 329 do if (I->data != TerminalState::BLANK) { } for huge performance gains :( Jan 24 23:01:38 it's comparable to the FBO object stuff I've been doing =/ Jan 24 23:01:58 s/object// Jan 24 23:08:42 you have to draw blanks Jan 24 23:08:47 for the underline Jan 24 23:08:55 also for the background Jan 24 23:12:29 not for current code you don't Jan 24 23:12:32 the screen gets cleared Jan 24 23:12:45 the underlines might be true though Jan 24 23:15:01 anyway, that's where I think most of our recent cpu loss is Jan 24 23:15:40 unless I'm mistaken we used to have variable line widths but now we pad with blanks? Jan 24 23:17:24 each char can have a different background colr Jan 24 23:17:30 you have to draw each of them Jan 24 23:18:53 never used gles2... and my opengl foo is a bit rusty anyway Jan 24 23:37:26 PuffTheMagic: i don't see any syslog messages, do i have to enable something? Jan 24 23:37:38 log level define Jan 24 23:38:19 stbuehler, export LOGLEVEL=7 Jan 24 23:38:21 for max output Jan 24 23:38:43 ah Jan 24 23:39:19 hm Jan 24 23:39:21 still nothing Jan 24 23:39:34 ah Jan 24 23:39:37 make clean.. Jan 24 23:39:38 make clean ; make LOGLEVEL=7 DEVICE=1 Jan 24 23:40:13 make -C src/plugin clean; make DEBUG=3 DEVICE=1 LOGLEVEL=7 && palm-install ipk/us.ryanhope.wterm_*.ipk && palm-launch us.ryanhope.wterm && palm-log -f us.ryanhope.wterm Jan 24 23:41:06 I don't think syslog stuff shows up in palm-log Jan 24 23:41:28 tail -f /var/log/messages | grep -o "wterm.plugin.*" Jan 24 23:41:59 strace -ff -p $(pidof pmsyslogd) -e trace=recv -s 4096 worked too :D Jan 24 23:42:14 lol Jan 24 23:42:55 imho LOG_ERR should reach syslog by default Jan 24 23:43:00 stbuehler, how to do test unicode chars from within wterm Jan 24 23:43:09 stbuehler, then change the default Jan 24 23:43:17 PuffTheMagic: select german layout :) Jan 24 23:43:45 that wont help Jan 24 23:44:05 i've added a new range in unicode_slots Jan 24 23:44:08 and want to test it Jan 24 23:44:23 i could try to convert it manually to utf-8 Jan 24 23:44:30 which unicode char? Jan 24 23:44:46 x0700 Jan 24 23:46:24 echo -e '\xBA\x80' Jan 24 23:47:28 no wait Jan 24 23:47:57 echo -e '\xDC\x80' Jan 24 23:48:40 up to \xDC\xBF, and \xDD\x80-\xDD\xBF Jan 24 23:48:57 i just get a box Jan 24 23:49:49 lame Jan 24 23:49:49 did you check whether the font supports it? Jan 24 23:49:59 i added it to the font ;) Jan 24 23:50:12 hm, perhaps my utf-8 magic sucks :D Jan 24 23:50:25 i found a ttf font with matrix chars Jan 24 23:50:37 you could just map some other char to 0x0700 in the sdl keyboard output, just set unicode = 0x0700 Jan 24 23:50:42 liberation mono Jan 25 00:16:20 stbuehler, i somehow need the ability to read a ttf font into a 128 byte slot Jan 25 00:16:33 and set it to one of the valid char slots Jan 25 00:17:04 into character set 2 actually Jan 25 00:17:25 or alternate 1 Jan 25 00:17:51 well, you can add a mapping in applyCharst Jan 25 00:18:48 then you can map ascii characters to some unicode values Jan 25 00:19:20 i need a more generic way Jan 25 00:19:50 in that case you have to be more precise what you want :) Jan 25 00:21:09 but somehow it sounds very bad to have different fonts loaded Jan 25 00:21:10 i want setfont to work ;) Jan 25 00:26:14 well obviouslly setfont cant work Jan 25 00:26:18 but i want something like it Jan 25 00:30:43 i dont even understand how applyCharset is working Jan 25 00:31:35 ahh Jan 25 00:31:38 I sorta get it Jan 25 00:34:07 if a certain charset is active it maps some (ascii) chars to a real unicode representation Jan 25 00:34:58 reading kde konsole source really helped me :) Jan 25 00:42:03 :D Jan 25 00:42:37 ok so how do I shift a char by 0x0700 Jan 25 00:54:56 case '1': if (c >= 0x20 && c <= 0x7f) c += 0x700 - 0x20; break; (or similar) Jan 25 01:00:02 i got it Jan 25 01:04:41 Brybry, if you have keycodes for the keyboard to emulate things we need I"d like to create a "help scene" for wTerm. Jan 25 01:08:56 i fail at fontforge Jan 25 01:10:58 hrmpf Jan 25 01:11:08 it fails to even compile a simple vertex shader Jan 25 01:11:19 void main() { gl_Position = vec4(1.0,1.0,1.0,1.0); } Jan 25 01:11:26 no error message ofc Jan 25 01:12:11 you guys need to pull from build.git Jan 25 01:12:14 and run make toolchain Jan 25 01:12:19 if u want to build latest wterm Jan 25 01:12:26