2025-02-18 00:53:30 <veltas      > crc: Well I think redraw() is the slow bit (although some lag is caused by waiting for KEYUP for input)
2025-02-18 00:53:44 <veltas      > But I too am struggling to sort out these performance issues
2025-02-18 00:54:08 <veltas      > I tried rewriting with SDL_Surface because that's supposedly better for pixel performance
2025-02-18 00:54:51 <veltas      > It seems like there's more than a couple issues with SDL's bitmap performance, and the docs aren't a lot of help trying to do the right thing
2025-02-18 00:55:22 <veltas      > Ironically 3D is really easy, but 2D pixel stuff is really slow
2025-02-18 00:56:39 <veltas      > It could be forced VSYNC, in which case redraw() might do better being limited to once a frame
2025-02-18 05:03:59 <MrMobius    > are you writing directly to memory with a pointer?
2025-02-18 05:04:26 <MrMobius    > I think I ended up using a texture for this. there are multiple ways to so it and a huge speed difference between them
2025-02-18 06:36:52 <ghodawalaaman> !gforth 1 1 + .
2025-02-18 06:36:52 <gforth_eval_bot> 2
2025-02-18 06:37:19 <ghodawalaaman> crc: I have made this simple bot, however it's still in work in progress
2025-02-18 06:37:31 <ghodawalaaman> It's just a illustration of how the bot will work
2025-02-18 06:38:00 <ghodawalaaman> you can play with the bot :)
2025-02-18 06:38:44 <ghodawalaaman> !gforth : SQUARE dup * ; 5 SQUARE .
2025-02-18 06:38:44 <gforth_eval_bot> 25
2025-02-18 06:38:53 <ghodawalaaman> !gforth : SQUARE dup * ; 10 SQUARE .
2025-02-18 06:38:54 <gforth_eval_bot> 100
2025-02-18 06:39:35 <ghodawalaaman> the bot is running inside a docker container
2025-02-18 08:50:52 <veltas      > MrMobius: If you've got any code for this somewhere and could dig it out it would be appreciated
2025-02-18 08:51:21 <veltas      > But I do suspect one issue is that redraw() which converts the VRAM into a surface or texture and then copies it to screen; is being called more often than it needs to be
2025-02-18 08:51:44 <veltas      > Once for every 'update', like every time a character is drawn, rather than once per display refresh
2025-02-18 08:53:28 <veltas      > And so there may be some unrequested vsync fuckery, or it could simply be a lot faster to just replace refresh() with refresh_next=1 and put refresh() in the main loop to only occur if (refresh_next && timer() > REFRESH_PERIOD)
2025-02-18 08:55:50 <veltas      > !gforth s" ls /" system
2025-02-18 08:55:50 <gforth_eval_bot> bin
2025-02-18 08:55:56 <veltas      > !gforth bye
2025-02-18 08:56:03 <veltas      > !gforth 5 .
2025-02-18 08:56:04 <gforth_eval_bot> 5
2025-02-18 08:56:29 <ghodawalaaman> veltas: it's running inside docker container
2025-02-18 08:56:42 <veltas      > Yeah I saw that
2025-02-18 08:57:10 <ghodawalaaman> it's only priting the first line of the output
2025-02-18 08:57:15 <ghodawalaaman> I have to make some changes in the code
2025-02-18 08:57:33 <ghodawalaaman> the bot is currently in alpha, I just wanted people to experiment with it
2025-02-18 08:57:53 <veltas      > Is the network part running outside of docker?
2025-02-18 08:58:37 <veltas      > !gforth s" which gforth" system
2025-02-18 08:58:38 <gforth_eval_bot> /usr/bin/gforth
2025-02-18 08:58:55 <ghodawalaaman> veltas: wait I forgot to disable internet access of the docker
2025-02-18 08:59:11 <veltas      > Might want to do that
2025-02-18 09:00:05 <veltas      > I'm not actually going to try and pwn you, but whenever someone puts a bot in IRC I will try and discover the issues with it because they quite often have huge security holes
2025-02-18 09:00:36 <ferret      > veltas: i am the most unhackable person on earth
2025-02-18 09:00:54 <ferret      > thanks to XEN lol
2025-02-18 09:05:49 <veltas      > !gforth s" ls / | fold -w" system
2025-02-18 09:12:13 <veltas      > What's unfortunate is that the VRAM for ilo-x is *almost* an SDL bitmap format, it's just going to require byteswapping because it's most-significant-bit and we can't assume a big endian platform
2025-02-18 09:12:40 <veltas      > But I don't think the byteswapping and blitting I'm doing slows it down much at all
2025-02-18 09:13:18 <ghodawalaaman> 
ACTION is back :)

2025-02-18 09:13:51 <ghodawalaaman> veltas: yes, my friend found a lot of issues with it and we fixed it
2025-02-18 09:13:57 <veltas      > Nice
2025-02-18 09:14:17 <veltas      > !gforth s" ls / | tr '\n' ''" system
2025-02-18 09:14:40 <veltas      > Are pipes just not allowed or am I bad at this
2025-02-18 09:14:41 <ghodawalaaman> !gforth s" ls " system
2025-02-18 09:15:04 <ghodawalaaman> tr: when not truncating set1, string2 must be non-empty
2025-02-18 09:15:04 <veltas      > Could be both
2025-02-18 09:15:11 <ghodawalaaman> Try 'fold --help' for more information.
2025-02-18 09:15:17 <ghodawalaaman> this is the error log I see
2025-02-18 09:15:21 <veltas      > !gforth s" ls / | tr '\n' ' '" system
2025-02-18 09:15:21 <gforth_eval_bot> bi
2025-02-18 09:15:35 <veltas      > !gforth s" ls / | tr '\n' ' '" system
2025-02-18 09:15:36 <gforth_eval_bot> bi
2025-02-18 09:15:40 <vms14       > :0 a forth bot
2025-02-18 09:15:51 <vms14       > !gforth ." :D" cr
2025-02-18 09:15:52 <gforth_eval_bot> :D
2025-02-18 09:16:15 <veltas      > !gforth s" mount" system
2025-02-18 09:16:15 <gforth_eval_bot> overlay on / type overlay (rw,relatime,lowerdir=/var/lib/docker/overlay2/l/4LWISPM42NNOCWYCC7IUF7ZYPY:/var/lib/docker/overlay2/l/XAVLZGBSSYTD6WFZY742GETCOZ:/var/lib/docker/overlay2/l/S2ZTY6O5GF6FT3UCBWZS3VQ44Z:/var/lib/docker/overlay2/l/WIBSR5UVBO4VZ4VNZV4YYZ3FG4:/var/lib/docker/overlay2/l/KSLMI4CNZO2BCOA3HPANSOQU63:/var/lib/docker/overlay2/l/LSEW6SGPU75NJXWBP7C3HT3NCS,upperdir=/var/lib/docker/overlay2/0a8c5a43adb
2025-02-18 09:16:31 <vms14       > oh it can flood
2025-02-18 09:16:44 <ghodawalaaman> vms14: no it only prints the first line of the output
2025-02-18 09:16:47 <vms14       > !gforth cr cr cr
2025-02-18 09:17:01 <xn435       > !gforth s" gforth" environment? [IF] type [THEN]
2025-02-18 09:17:02 <gforth_eval_bot> 0.7.3
2025-02-18 09:17:03 <vms14       > what if the first line is large
2025-02-18 09:17:26 <veltas      > Then it prints under 512 bytes
2025-02-18 09:17:38 <veltas      > It cut it off earlier
2025-02-18 09:17:38 <vms14       > then it's fine
2025-02-18 09:18:29 <vms14       > ghodawalaaman it's cool that you made this bot
2025-02-18 09:18:47 <ghodawalaaman> Thank you :3
2025-02-18 09:19:04 <vms14       > it will be handy when discussing anything about forth
2025-02-18 09:19:08 <veltas      > Yes it is great ghodawalaaman, this will be very useful if it stays up and doesn't get you hacked
2025-02-18 09:19:41 <xn435       > I'm afraid to try [BEGIN] ." asdf" [AGAIN]
2025-02-18 09:19:47 <ghodawalaaman> there is currently a vulnerability that let's you execute arbitary IRC command from the bot
2025-02-18 09:20:06 <ghodawalaaman> xn435: the command will timeout
2025-02-18 09:20:08 <veltas      > You might want to ask pragma in #c what he did to secure his compiler bots
2025-02-18 09:20:29 <vms14       > pragma is a perl dev
2025-02-18 09:20:31 <xn435       > !gforth ." /nickserv"
2025-02-18 09:20:31 <gforth_eval_bot> /nickserv
2025-02-18 09:20:54 <ghodawalaaman> xn435: the second line will be executed as an IRC command
2025-02-18 09:21:02 <ghodawalaaman> don't do something malicious with it please
2025-02-18 09:21:06 <xn435       > !gforth cr ." /nickserv"
2025-02-18 09:21:09 <ghodawalaaman> veltas: I will ask him thank you!
2025-02-18 09:21:21 <xn435       > nah I don't intend to :)
2025-02-18 09:21:23 <xn435       > nice bot
2025-02-18 09:21:35 <ghodawalaaman> :3
2025-02-18 09:22:30 <veltas      > vms14: He's got it running GCC so it's definitely sandboxed
2025-02-18 09:22:44 <veltas      > And given what #c is like I'm assuming it's quite secure by now
2025-02-18 09:22:59 <vms14       > the perlbot is also sandboxed
2025-02-18 09:23:18 <vms14       > you cannot mess with it
2025-02-18 09:24:35 <vms14       > I made one for my toy lang once, but you could rm -rf my system xd
2025-02-18 09:25:41 <vms14       > didn't bother to prevent anything, but in my case I could remove harmful words
2025-02-18 09:27:54 <vms14       > ghodawalaaman I assume you made a restricted user for the bot
2025-02-18 09:28:45 <ghodawalaaman> right now I haven't, I have to fix a lot of issue/vulnerability yet, but thank you for the suggesting
2025-02-18 09:28:56 <ghodawalaaman> right now it's more like proof of concept type bot
2025-02-18 09:29:30 <vms14       > this channel is small so there is less chance that someone will mess with it
2025-02-18 09:29:50 <vms14       > but it can happen anyways and you should be aware
2025-02-18 09:30:55 <vms14       > for now you could try to fix the most dangerous problems until you learn how to improve it
2025-02-18 09:31:23 <ghodawalaaman> yeah, I will make it more usable and secure!
2025-02-18 09:31:29 <vms14       > if you make a user you can mess with the path env variable so it won't find the commands like ls
2025-02-18 09:31:38 <ghodawalaaman> hopefully till weekend I will have a good version of the bot
2025-02-18 09:31:43 <vms14       > still io is a problem
2025-02-18 09:32:32 <ghodawalaaman> vms14: changing path variable is a neat idea
2025-02-18 09:32:42 <vms14       > but if the user has no write permissions then it's fine
2025-02-18 09:33:02 <ghodawalaaman> yeah
2025-02-18 09:33:57 <vms14       > you can also redefine some forth words
2025-02-18 09:34:58 <vms14       > !gforth : system ." :D" cr ;
2025-02-18 09:35:28 <xn435       > !gforth s" unix/socket.fs" included 2048 BUFFER: buf s" example.com" 80 open-socket dup s\" GET /\r\n\r\n" rot write-socket 500 ms dup buf 2048 read-socket type close-socket
2025-02-18 09:35:29 <vms14       > !gforth s" ls" system
2025-02-18 09:35:50 <vms14       > oh forgot the bot was offline lol
2025-02-18 09:38:19 <ghodawalaaman> xn435: I forgot to disable interenet of the docker container
2025-02-18 09:39:04 <ghodawalaaman> internet*
2025-02-18 09:39:17 <xn435       > yeah that could be an issue
2025-02-18 09:39:22 <ghodawalaaman> vms14: I will connect it again after I fix the issues :)
2025-02-18 09:41:53 <xn435       > I'm wondering if creating a custom wordlist and limiting memory reads and writes would be enough to secure it
2025-02-18 09:42:21 <xn435       > that is, only allowing to write and read from specific ranges
2025-02-18 09:43:07 <vms14       > it also has to prevent infinite loops, like some kind of watcher that will kill the process after a timeout
2025-02-18 09:43:29 <xn435       > yeah he mentioned that it would time out
2025-02-18 09:43:56 <vms14       > but there is always someone smarter that finds a way
2025-02-18 09:54:32 <veltas      > I had considered doing a Forth VM specifically designed to be run as a sandbox, e.g. for IRC or whatever situation it needed to be safe
2025-02-18 09:54:44 <veltas      > You can adapt existing VM's, e.g. ilo, and make them safe
2025-02-18 09:55:01 <veltas      > ilo is already nearly there, it just needs memory bounds checking really
2025-02-18 09:55:21 <veltas      > So doing a sandboxed konilo bot would be quite easy
2025-02-18 09:56:23 <veltas      > But we need an ANS Forth, Forth 2012 etc because most of our questions are about classic/standard forths
2025-02-18 09:57:13 <veltas      > The other idea was to treat the VM like a sort of old mainframe, and allow people access to private storage as well as shared storage for communicating with each other
2025-02-18 09:57:19 <veltas      > And sharing programs in blocks
2025-02-18 09:57:55 <veltas      > People could define their own command aliases in blocks
2025-02-18 09:58:24 <veltas      > There's a ton of fun stuff that can be done here, we forget sometimes that the internet is quite fun
2025-02-18 10:01:41 <pgimeno     > ghodawalaaman: this is a must read for people who want to make their bot secure: https://irc-bot-science.clsr.net/
2025-02-18 10:01:58 <pgimeno     > and on top of that there's securing the interpreter itself of course
2025-02-18 10:03:12 <pgimeno     > well that's mainly for parsing links
2025-02-18 10:07:35 <pgimeno     > but if it's going to output user-generated text, it may give you ideas of things to prevent (like certain chars)
2025-02-18 13:01:44 <MrMobius    > veltas: https://pastebin.com/RveGnqXD
2025-02-18 13:14:38 <crc         > veltas: re: bounds checking; see http://forth.works/ilo-unix.c for a copy of ilo with this
2025-02-18 13:22:34 <veltas      > MrMobius: Try 4 is actually what crc was using, except for STREAMING option which I tried and it made no difference
2025-02-18 13:22:44 <veltas      > I wonder if there's a difference in drivers affecting things here
2025-02-18 13:24:45 <veltas      > Assuming try 4 is the source shown
2025-02-18 13:25:03 <MrMobius    > yes
2025-02-18 13:25:17 <MrMobius    > see line 27
2025-02-18 13:26:47 <veltas      > I'll try delaying to refresh rate tonight probably, with the streaming texture
2025-02-18 13:27:08 <veltas      > And if that works well enough I'll fix the input lag too
2025-02-18 13:27:19 <veltas      > Or anyone else is welcome to give it a go
2025-02-18 13:28:14 <MrMobius    > what's the issue?
2025-02-18 13:31:16 <veltas      > ilo-sdl is slow basically
2025-02-18 13:32:05 <veltas      > You can try building it and run, and stuff like `catalogue` prints very slowly
2025-02-18 13:33:03 <ghodawalaaman> pgimeno: Thank you for the link!
2025-02-18 15:22:57 <ghodawalaaman> I have added the bot again
2025-02-18 15:23:25 <ghodawalaaman> however It still has some vulnerability which I couldn't fix
2025-02-18 15:23:59 <ghodawalaaman> gforth_eval_bot will work fine unless someone actually trying to crash the bot
2025-02-18 15:24:19 <ghodawalaaman> !gforth 1 1 cr . cr .
2025-02-18 15:24:22 <gforth_eval_bot> https://0x0.st/8beJ.txt
2025-02-18 15:24:28 <ghodawalaaman> !gforth 1 1 +
2025-02-18 15:24:49 <ghodawalaaman> !gforth 1 1 +
2025-02-18 15:25:26 <ghodawalaaman> !gforth 1 1 + .
2025-02-18 15:25:27 <gforth_eval_bot> 2
2025-02-18 15:25:40 <ghodawalaaman> !gforth 1 1 . cr .
2025-02-18 15:25:41 <gforth_eval_bot> 1
2025-02-18 15:25:43 <gforth_eval_bot> https://0x0.st/8bey.txt
2025-02-18 15:25:51 <ghodawalaaman> It also have support for multiline output now
2025-02-18 15:26:50 <ghodawalaaman> s/have/has
2025-02-18 15:42:00 <dzho        > !gforth 42 EMIT
2025-02-18 15:42:00 <gforth_eval_bot> *
2025-02-18 15:42:26 <dzho        > !gforth : tostar 42 emit ; tostar
2025-02-18 15:42:27 <gforth_eval_bot> *
2025-02-18 15:42:39 <dzho        > !gforth tostar
2025-02-18 15:42:52 <dzho        > sorry
2025-02-18 15:43:15 <ghodawalaaman> dzho: it's ok, you can test it now :)
2025-02-18 15:43:51 <ghodawalaaman> !gforth tostar
2025-02-18 15:43:59 <ghodawalaaman> !gforth 1 1 + .
2025-02-18 15:44:14 <dzho        > !gforth : tostar 42 emit ; 91 emit tostar
2025-02-18 15:44:19 <dzho        > derp
2025-02-18 15:44:26 <ghodawalaaman> I don't know why it's disconnecting
2025-02-18 15:44:31 <ghodawalaaman> smh
2025-02-18 15:44:35 <dzho        > !gforth 91 emit
2025-02-18 15:44:35 <gforth_eval_bot> [
2025-02-18 15:44:50 <dzho        > !gforth 89 emit
2025-02-18 15:44:51 <gforth_eval_bot> Y
2025-02-18 15:44:59 <dzho        > !gforth : tostar 42 emit ; 89 emit tostar
2025-02-18 15:45:00 <gforth_eval_bot> Y*
2025-02-18 15:45:11 <dzho        > ok maybe I'm just picking horrible example characters
2025-02-18 16:21:20 <ghodawalaaman> !gforth : x 1000 0 DO I . LOOP ; x
2025-02-18 16:21:21 <gforth_eval_bot> 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
2025-02-18 16:21:23 <gforth_eval_bot> https://0x0.st/8b21.txt
2025-02-18 16:21:37 <ghodawalaaman> all the vulnerabilities are fixed now
2025-02-18 16:22:08 <ghodawalaaman> I will wait 1 day, if no bugs are discovered I will host it on my VPS
2025-02-18 16:30:53 <user51      > !gforth : r recurse ; r
2025-02-18 16:30:53 <ghodawalaaman> *OS command line*:-1: Return stack overflow
2025-02-18 16:31:04 <ghodawalaaman> maybe I should print the error messages too
2025-02-18 16:31:04 <user51      > that's the nice version.. now for the evil version
2025-02-18 16:31:11 <user51      > : r rdrop recurse ; r
2025-02-18 16:31:18 <user51      > !gforth : r rdrop recurse ; r
2025-02-18 16:32:29 <ghodawalaaman> user51: the bot is still fine
2025-02-18 16:32:57 <ghodawalaaman> !gforth : R dup * ; 1223 R
2025-02-18 16:33:07 <user51      > ghodawalaaman: well done.  :)
2025-02-18 16:33:21 <ghodawalaaman> !gforth : R dup * ; 1223 R .
2025-02-18 16:33:22 <gforth_eval_bot> 1495729
2025-02-18 16:33:26 <user51      > !gforth rdrop
2025-02-18 16:33:55 <ghodawalaaman> the bot won't remeber any WORD definitions
2025-02-18 16:34:11 <ghodawalaaman> bot launches a new instance of gforth everytime
2025-02-18 16:35:40 <veltas      > "All the vulnerabilities are fixed now" bold claim lol
2025-02-18 16:35:52 <veltas      > Nice job ghodawalaaman
2025-02-18 16:37:00 <user51      > ghodawalaaman: maybe it should accept PMs
2025-02-18 16:37:10 <user51      > also, my final test
2025-02-18 16:37:18 <user51      > !gforth words
2025-02-18 16:37:21 <gforth_eval_bot> https://0x0.st/8b2g.txt
2025-02-18 16:38:11 <veltas      > !gforth : x 0 0 do i . loop ; x
2025-02-18 16:38:16 <gforth_eval_bot> 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
2025-02-18 16:38:19 <gforth_eval_bot> https://0x0.st/8b2E.txt
2025-02-18 16:46:38 <ghodawalaaman> veltas: Thank you!
2025-02-18 16:47:13 <ghodawalaaman> user51: yeah, maybe but coding that is tricky
2025-02-18 21:32:17 <KipIngram   > Well, I got kind of revved up on the idea of a Forth processor again yesterday and today.  Had a couple of thoughts about it that were new for me, and it perked my interest again.  Been eyeing my Artix A7 kit.  :-)
2025-02-18 22:07:09 <veltas      > crc: https://pastebin.com/raw/Luw4qYTc
2025-02-18 22:07:52 <veltas      > For whatever reason this patch has vastly improved performance for me, so hopefully that helps you too
2025-02-18 22:08:04 <veltas      > I changed to STREAMING in there but not sure if that even makes any difference
2025-02-18 22:09:21 <veltas      > Well it probably makes difference in certain scenarios, but I don't think that was the limiting factor
2025-02-18 22:49:09 <veltas      > If you just OR SDL_WINDOW_SHOWN with SDL_WINDOW_RESIZABLE then the window is resizable
2025-02-18 22:49:23 <veltas      > And just do_redraw=1 on any SDL_WINDOWEVENT
2025-02-18 22:49:49 <veltas      > So that it actually redraws if you unhide or resize the window
2025-02-18 22:50:00 <vms14       > veltas what are you doing with sdl?
2025-02-18 22:50:06 <vms14       > some app or a game
2025-02-18 22:50:51 <veltas      > What am I not doing with it
2025-02-18 22:50:51 <vms14       > :0
2025-02-18 22:51:10 <vms14       > I only played a bit with xlib, but it was very cool
2025-02-18 22:51:21 <vms14       > I made a charmander walking on the screen with xshape
2025-02-18 22:51:30 <veltas      > crc previously tried SDL and ran into performance issues which I think I've fixed (at least for me I have, I don't know if it will fix for him too)
2025-02-18 22:51:43 <vms14       > it was a non rectangular window changing it's shape with the frames of the sprite
2025-02-18 22:52:32 <veltas      > I wanted SDL because I wanted a resizable window, and didn't know how to do it for X
2025-02-18 22:55:40 <veltas      > X Is Applicable For Point-Of-Sale?
2025-02-18 22:55:43 <vms14       > ConfigureNotify events
2025-02-18 22:56:27 <vms14       > do you mean if it makes sense to make a point of sale app with xlib?
2025-02-18 22:56:31 <veltas      > I also like that SDL is cross platform
2025-02-18 22:56:34 <vms14       > if you mean that, then no
2025-02-18 22:56:50 <veltas      > No don't worry vms14, making a joke about xentrac's filename
2025-02-18 22:56:58 <vms14       > ah lol
2025-02-18 22:59:41 <xentrac     > veltas: X is a f****** piece of s***
2025-02-18 23:00:20 <vms14       > the funny thing is that wayland is not much better xd
2025-02-18 23:08:36 <GreaseMonkey> handling window resizing in X? there's 2 events, and one of them is definitely not the one you want
2025-02-18 23:09:55 <GreaseMonkey> the one you don't want is the one with "Resize" in its name, specifically "ResizeRequest"
2025-02-18 23:11:25 <GreaseMonkey> i *think* you want something like ConfigureNotify?
2025-02-18 23:11:54 <veltas      > Classic IRC, I said I don't want to use X, and everyone tells me how to use X
2025-02-18 23:12:07 <veltas      > Oh well at least I learned something about tech I never want to use
2025-02-18 23:13:11 <veltas      > Is that because ResizeRequest is the *request*, and the resize triggers a ConfigureNotify?  And if you intercept the ResizeRequest then it never happens?
2025-02-18 23:14:47 <veltas      > This reminds me of win32.  They're very similar, except win32 isn't always a network protocol
2025-02-18 23:24:01 <xentrac     > that sounds right
2025-02-18 23:31:31 <veltas      > crc: Fixed keyboard/input issues as well https://pastebin.com/raw/yyf4guyW
2025-02-18 23:46:27 <veltas      > Okay and I've also stopped it from using 100% CPU all the time, and removed debug keyboard queue prints https://pastebin.com/raw/rZsHsYgR
2025-02-18 23:56:38 <veltas      > Looks like the X demo crashes so it probably still needs some stuff added
2025-02-18 23:57:13 <veltas      > #653 I mean