| Time |
Nick |
Message |
| 01:35 |
Parnikkapore_m |
<independent56> "And yet there are people who use..." <- *knocks on real life cars* |
| 01:38 |
Parnikkapore_m |
tbf it is useful in cases where you can't cluster trips into lines |
| 01:41 |
Parnikkapore_m |
I find building rail funner, but there's also bike infra (but to some extent it's a scaled down version of car things) running mostly parallel and to some low-traffic places |
| 01:43 |
|
Soni joined #minetest |
| 01:43 |
Parnikkapore_m |
which makes my pillars of transport / transport hierarchy go walking, cycling, L/R/E trains |
| 01:45 |
ball |
What's an L/R/E? |
| 01:45 |
ball |
Local/Regional/Express? |
| 01:47 |
Parnikkapore_m |
yup |
| 01:47 |
ball |
It would be nice to live in a country with practical intercity rail service. |
| 01:47 |
Parnikkapore_m |
way too many categories for Minetest but that's part of the fun |
| 01:47 |
muurkha |
hmm, I've had an idea for a while about real-world automated light rapid transit in tunnels |
| 01:48 |
muurkha |
PRT type |
| 01:48 |
muurkha |
maybe I should build a demo |
| 01:48 |
muurkha |
in Minetest |
| 01:48 |
Parnikkapore_m |
^ as someone who lives in a place where intercity rail is a last resort, I agree |
| 01:49 |
ball |
muurkha: Tunnels are expensive. |
| 01:49 |
muurkha |
basically the idea is that cities would be a lot better in a lot of ways with better transport |
| 01:49 |
Parnikkapore_m |
muurkha: I made one https://parnikkapore.neocities.org/projects/mtprt.htm |
| 01:49 |
muurkha |
yes, they are, but they get much cheaper when you make them much smaller, ball |
| 01:49 |
Parnikkapore_m |
surface corridors tho since the actual infra isn't part of the project |
| 01:49 |
muurkha |
my theory is that if you recline when you get into the vehicle then you can make the tunnel only about two meters tall |
| 01:49 |
Parnikkapore_m |
cue "the Loop should be either a controlled access highway or a PRT" |
| 01:50 |
ball |
muurkha: You'd like the Glasgow subway then. |
| 01:50 |
muurkha |
if the riders are one behind the other then you can make it only about two meters wide |
| 01:50 |
Parnikkapore_m |
yee check out glasgow subway |
| 01:50 |
muurkha |
and if the whole vehicle only weighs about 400 kg then you can use dirt floors |
| 01:51 |
muurkha |
no rails, no concrete floors, no concrete walls, just concrete ceilings |
| 01:52 |
Parnikkapore_m |
not sure if concrete ceilings only would work |
| 01:52 |
* ball |
looks confused |
| 01:52 |
muurkha |
if you reserve the whole path from beginning to end when you start the journey, you don't have to worry about a vehicle in front of you stopping and you crashing into it |
| 01:52 |
Parnikkapore_m |
rl geotechnical engineering is complex |
| 01:52 |
Parnikkapore_m |
muurkha: sounds like a terribly inefficient system |
| 01:52 |
muurkha |
well, it has the usual efficiencies of PRT |
| 01:53 |
muurkha |
and by virtue of having tunnels that are an order of magnitude smaller and two orders of magnitude cheaper than what you need for even light rail, it costs enormously less per mile |
| 01:54 |
ball |
I'm not convinced the cost of boring a tunnel decreases significantly with diameter |
| 01:54 |
muurkha |
and because the path is clear from start to destination, you can go at much higher speeds |
| 01:54 |
muurkha |
yeah, ideally you want to trench and cover, not bore |
| 01:55 |
ball |
Some of the London Underground was built cut-and-cover |
| 01:55 |
ball |
These days a subway has to be built further down, I think. |
| 01:56 |
muurkha |
well, my utopian aspirations don't end there |
| 01:56 |
ball |
You'd probably enjoy Jago Hazzard's channel on the YouTube. |
| 01:56 |
ball |
I know I do. |
| 01:57 |
muurkha |
I want to keep cars out of the city entirely, so what's running over the top of the tunnels is just parkland with people picnicking |
| 01:57 |
muurkha |
not trucks and honking taxis |
| 01:57 |
ball |
Good luck with that. |
| 01:57 |
muurkha |
my notes from three years ago are in https://dercuano.github.io/notes/tunnel-prt.html |
| 01:59 |
muurkha |
I was calculating with 1.5-meter diameter, but in Minetest I'd have to use 2-meter |
| 01:59 |
* ball |
sighs |
| 01:59 |
ball |
Time to reboot my wireless bridge again. |
| 02:00 |
muurkha |
you could get anywhere within a 4-km radius in 5 minutes: 50 square kilometers or 5000 city blocks |
| 02:00 |
muurkha |
plus half a minute on each end of the trip to walk to or from the station |
| 02:02 |
muurkha |
Parnikkapore_m: your mod looks super cool! |
| 02:03 |
muurkha |
I'm curious why you didn't go with keeping track of where each car is and picking out a car when someone calls it |
| 02:03 |
Parnikkapore_m |
it's a build using advtrains and luaatc ? |
| 02:04 |
Parnikkapore_m |
muurkha: Mostly for complexity reasons. I also don't see the advantage other than knowing the arrival time of the empty car |
| 02:05 |
muurkha |
yeah, I guess in Minetest you don't have to worry about empty cars running over people |
| 02:06 |
Parnikkapore_m |
Nor energy efficiency |
| 02:06 |
Parnikkapore_m |
A PRT like that should be grade separated anyway |
| 02:06 |
muurkha |
nor wear |
| 02:07 |
muurkha |
yeah, like I said, tunnels are my favored solution there |
| 02:07 |
muurkha |
without rails! |
| 02:08 |
Parnikkapore_m |
anw my main concern with dirt tunnels is about soil pressure causing the ground to bulge or worse |
| 02:09 |
muurkha |
yeah, that's addressed in https://dercuano.github.io/notes/tunnel-prt.html#addtoc_3, isn't it? |
| 02:09 |
muurkha |
or did you mean something else? |
| 02:14 |
Parnikkapore_m |
it's less about shear failure and more about the ground pushing against the tunnel walls |
| 02:20 |
muurkha |
oh, yeah, you do need to stabilize the walls. do you know about mechanically stabilized earth? |
| 02:22 |
Parnikkapore_m |
yup |
| 02:22 |
Parnikkapore_m |
but I'm not sure if it'll help |
| 02:23 |
Parnikkapore_m |
remember we're not dealing with load bearing capacity / shear failure here |
| 02:24 |
muurkha |
do you mean compressive failure, tensile failure, creep...? |
| 02:26 |
muurkha |
I'm no expert on geoengineering so I could totally be missing something basic |
| 02:54 |
|
wsor joined #minetest |
| 02:54 |
|
Edgy2 joined #minetest |
| 03:47 |
|
lemonzest joined #minetest |
| 04:00 |
|
MTDiscord joined #minetest |
| 05:04 |
|
ball left #minetest |
| 05:23 |
|
calcul0n_ joined #minetest |
| 06:54 |
|
ronoaldo joined #minetest |
| 07:08 |
|
orwell96 joined #minetest |
| 08:54 |
|
___nick___ joined #minetest |
| 09:07 |
|
appguru joined #minetest |
| 09:17 |
|
Sven_vB joined #minetest |
| 10:03 |
|
behalebabo joined #minetest |
| 11:33 |
|
z812 joined #minetest |
| 11:35 |
|
Fixer joined #minetest |
| 11:48 |
|
specing_ joined #minetest |
| 12:34 |
|
definitelya joined #minetest |
| 12:44 |
|
debiankaios joined #minetest |
| 13:38 |
|
sec0nd joined #minetest |
| 14:14 |
|
lemonzest joined #minetest |
| 14:23 |
|
est31 joined #minetest |
| 14:24 |
|
Taoki joined #minetest |
| 15:11 |
|
cation joined #minetest |
| 15:32 |
|
debiankaios joined #minetest |
| 15:42 |
sfan5 |
!mod mineclone2 |
| 15:42 |
MinetestBot |
sfan5: Quick Harvest & Replant for MineClone2 / MineClone5 / Mineclonia [mcl_quick_harvest_replant] by erlehmann - https://forum.minetest.net/viewtopic.php?t=28104 - https://git.minetest.land/erlehmann/mcl_quick_harvest_replant |
| 15:42 |
sfan5 |
no... |
| 15:45 |
sfan5 |
rubenwardy: can you contribute a !cdb command? :D |
| 16:08 |
MTDiscord |
<Jonathon> Discord minetest bot > irc minetest bot |
| 16:27 |
|
Kilroy joined #minetest |
| 16:32 |
MinetestBot |
[git] sjml -> minetest/minetest: Fix permissions on workflow-generated macOS builds (#12422) c10fe7e https://github.com/minetest/minetest/commit/c10fe7ec18eed354a5b13c849198cede79041adc (2022-06-13T16:30:48Z) |
| 16:32 |
MinetestBot |
[git] sfan5 -> minetest/minetest: Improve lua_api.txt a bunch 804c255 https://github.com/minetest/minetest/commit/804c2559411fc1d03644a5eac2c37c7ea2ff79a7 (2022-06-13T16:31:03Z) |
| 16:32 |
MinetestBot |
[git] Zughy -> minetest/minetest: lua_api.txt: add missing backticks 0f3f1a0 https://github.com/minetest/minetest/commit/0f3f1a001c0dcaa91169bbc90900a47120bb2763 (2022-06-13T16:31:33Z) |
| 16:36 |
rubenwardy |
Do we have any stats on the language use of our users? |
| 16:36 |
rubenwardy |
I suppose Google Play is probably the extent of this, unless the server list gathers that info (which it probably shouldn't) |
| 16:54 |
|
Talkless joined #minetest |
| 16:55 |
MTDiscord |
<MNH48> even if the server list gathers that info, it might not be accurate anyway... since not everyone plays on servers that is listed in server list, some just play in LAN setting or private non-public Minetest servers |
| 16:56 |
MTDiscord |
<MNH48> also some who play singleplayer only |
| 17:12 |
|
___nick___ joined #minetest |
| 17:14 |
|
___nick___ joined #minetest |
| 17:19 |
|
Verticen_ joined #minetest |
| 17:34 |
|
sobkas joined #minetest |
| 18:14 |
|
debiankaios joined #minetest |
| 18:14 |
|
sobkas joined #minetest |
| 18:14 |
rubenwardy |
64% of Google Play users have don't have English as their primary language |
| 18:17 |
rubenwardy |
-> 12% Russian, 11,% Spanish, 9% German, 7% French, 4% portuguese |
| 18:17 |
muurkha |
of those who installed Minetest from it, or overall? |
| 18:17 |
rubenwardy |
Our current users |
| 18:17 |
muurkha |
I installed it on my Spanish cellphone, but not from Google Play |
| 18:18 |
rubenwardy |
if you don't have GPlay installed, then you won't be counted |
| 18:18 |
rubenwardy |
I believe it may still count you if you have GPlay installed but install from another source, but not sure |
| 18:18 |
muurkha |
I can't uninstall Google Play until and unless I root the phone |
| 18:18 |
muurkha |
unfortunately |
| 18:18 |
muurkha |
I could use the space |
| 18:18 |
muurkha |
it's interesting that Russian is so popular and Arabic, Hindi, and Chinese are so unpopular |
| 18:19 |
rubenwardy |
Anyway, I want to have a stat for number of non-English users to encourage modders to add translation support. 64% is compelling, even if not accurate |
| 18:19 |
muurkha |
yeah |
| 18:19 |
rubenwardy |
Chinese users will probably use another store |
| 18:19 |
muurkha |
I wish I could type things like "ñ" and "é" in the chat |
| 18:20 |
muurkha |
on desktop Linux, not Android |
| 18:20 |
rubenwardy |
version? That should be fixed in newer versions |
| 18:21 |
rubenwardy |
make sure you have at least 5.5.0, because that had a lot of keyboard and unicode fixes |
| 18:21 |
muurkha |
5.3.0 |
| 18:21 |
muurkha |
actually ñ works in 5.3.0 with a Spanish-language keyboard, but é etc. don't |
| 18:22 |
rubenwardy |
yeah, definitely update then |
| 18:22 |
sfan5 |
you can even type 具体的な単語 in chat if you so wanted in 5.5.0 |
| 18:22 |
sfan5 |
(Linux and Android only) |
| 18:22 |
rubenwardy |
2020 is 2 years old, 5.5.0 has loads of unicode fixes |
| 18:22 |
rubenwardy |
5.3.0 is 2 years old (2020), 5.5.0 has loads of unicode fixes |
| 18:22 |
sfan5 |
"2020 is 2 years old" technically true |
| 18:22 |
rubenwardy |
lol |
| 18:23 |
muurkha |
<2020> no! I'm 2½! |
| 18:23 |
muurkha |
that's sweet, I definitely want to type things like ☢ and ☣ in the chat |
| 18:24 |
rubenwardy |
people forget that unicode isn't just emoji |
| 18:24 |
muurkha |
well, I can hardly forget that speaking Spanish every day |
| 18:25 |
muurkha |
but I like the emojiish things too! |
| 18:26 |
muurkha |
maybe Russian is high and Arabic is low because of the fraction of ESL speakers in those languages — maybe English forums are still a memetically very important factor |
| 18:27 |
muurkha |
by itself that doesn't really explain Hindi though, nor Russian being higher than German |
| 18:29 |
MinetestBot |
[git] sfan5 -> minetest/minetest_game: Update engine feature checks 986b657 https://github.com/minetest/minetest_game/commit/986b65775f3d7a89959f569c516af8c44b4bc7e3 (2022-06-13T18:02:08Z) |
| 18:30 |
muurkha |
actually I guess that 2020 is 3 years old, if we consider it to have been born on Jan 1, 2020, rather than Dec 31, 2020 |
| 18:30 |
muurkha |
no, never mind, it's still 2½ |
| 18:32 |
|
sys4 joined #minetest |
| 18:32 |
|
Verticen_ joined #minetest |
| 18:34 |
|
Yad joined #minetest |
| 18:44 |
|
sobkas joined #minetest |
| 18:53 |
|
sys4 joined #minetest |
| 19:09 |
MTDiscord |
<MNH48> >it's interesting that Russian is so popular and Arabic, Hindi, and Chinese are so unpopular |
| 19:09 |
MTDiscord |
<MNH48> Hindi and Arabic both need complex script support (where the text will change depending on position/sequence of the character) |
| 19:09 |
MTDiscord |
<MNH48> which iirc is still not supported (so it will only display the base character instead) |
| 19:09 |
MTDiscord |
<MNH48> so probably the people using those languages just don't play Minetest because of that |
| 19:10 |
erle |
can you say how viable unifont is for these kinds of scripts? |
| 19:10 |
MTDiscord |
<MNH48> it's not about font, it's about rendering |
| 19:10 |
erle |
i know, but i wonder how viable a fixed-witth glyph setup is for this |
| 19:10 |
erle |
width |
| 19:10 |
MTDiscord |
<MNH48> like you can throw in supported font all you want, but if the rendering don't work it just don't |
| 19:11 |
erle |
rubenwardy sfan5 how is the situation for scripts that go right to left? |
| 19:11 |
rubenwardy |
poor |
| 19:11 |
rubenwardy |
we don't support RTL text currently |
| 19:12 |
erle |
*sad TV channel noises* |
| 19:12 |
|
sec0nd joined #minetest |
| 19:18 |
|
sobkas joined #minetest |
| 19:18 |
|
Noisytoot_ joined #minetest |
| 19:20 |
|
sobkas joined #minetest |
| 19:25 |
muurkha |
MNH48: thanks! |
| 19:26 |
muurkha |
erle: the issue is that for Arabic you have three or four different glyphs per character (code point) depending on whether it's at the beginning, middle, or end of a word, or by itself; and vowels are rendered as accents above or below the text |
| 19:26 |
muurkha |
also it's RTL |
| 19:27 |
muurkha |
Hindi is written in Devanagari, which is LTR, but the glyph for a syllable consists of the consonant with optionally a vowel sort of wrapped around it |
| 19:32 |
muurkha |
in both cases you can do fixed-width glyphs I think, though I haven't seen an example; the problem is just that the codepoint-sequence-to-glyph-sequence mapping is sort of complicated |
| 19:33 |
muurkha |
and although in principle you could do a set of mappings similar to the ones some terminals use for combining accents e + ’ → é, at least if the right side can contain multiple characters, that is a pretty unwieldy way to do it |
| 19:34 |
muurkha |
erle: btw you might be interested in this non-minetest-related thing I wrote last month: http://canonical.org/~kragen/sw/dev3/trama |
| 19:35 |
muurkha |
if you are the glitch URL erlehmann, apologies if that's a case of mistaken identity |
| 19:37 |
MTDiscord |
<MNH48> that's basically what "complex script" behave in unicode, the actual 'character' (just one codepoint) is rendered differently depending on its position or sequence |
| 19:51 |
erle |
muurkha yes i made libglitch |
| 19:51 |
erle |
muurkha, are you kragen? |
| 19:55 |
|
sobkas joined #minetest |
| 19:55 |
MTDiscord |
<MNH48> for example on Arabic script, see how the in-game display it even with a font that fully support the Arabic script: |
| 19:55 |
MTDiscord |
<MNH48> https://cdn.discordapp.com/attachments/749727888659447960/985995939556773898/2022-06-14_03-50-36_minetest.png |
| 19:56 |
MTDiscord |
<MNH48> versus what it actually is: |
| 19:56 |
MTDiscord |
<MNH48> https://cdn.discordapp.com/attachments/749727888659447960/985995976865108028/unknown.png |
| 19:56 |
MTDiscord |
<MNH48> plaintext: اين سواتو ڤرچوباٴن توليسن |
| 19:56 |
erle |
wait i make a screenshot |
| 19:58 |
rubenwardy |
how many possible users do we have that use a RTL language? |
| 19:58 |
rubenwardy |
According to a random website, Arabic speakers make up 4.8% of the Internet |
| 19:59 |
erle |
does this look correct? (hexchat, unifont) data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAANcAAAAWAQMAAABpB5vnAAAABlBMVEUAAAD///+l2Z/dAAAAqklEQVQoU9XOwWqEYAyF0WMIP1oK6uCACzeKD9H3fy5ruxi1DLSl214uCcnHDeEvev2FdaD+lv2QC+g6uCG5qjA4jzJ8sfqRqy92bK+cMzd5ullllESXhZuSWSoyEfeMJoQumrd40UQ2d5GJ/h7RhF4XTanKwfpIzB/vW7uZtXu7V5t2y3Y3b4kZ1dkqh2ewGAQLYggeXobz65WJerU6PDk1Hq0eGa/xv+gT7AcQIFuA/R0AAAAASUVORK5CYII= |
| 19:59 |
rubenwardy |
looks like a data url with base64 |
| 20:00 |
erle |
hand it to a web browser and you see how اين سواتو ڤرچوباٴن توليسن renders here |
| 20:00 |
MTDiscord |
<MNH48> (after using base64 to image converter) it does look correct |
| 20:00 |
erle |
what is a “base64 to image converter”? |
| 20:00 |
MTDiscord |
<MNH48> https://codebeautify.org/base64-to-image-converter |
| 20:01 |
MTDiscord |
<MNH48> because you sent base64 strings lol |
| 20:01 |
erle |
you can put that stuff right in the address bar of your browser or right click the link and open it or so if your url handler supports the data: scheme |
| 20:01 |
rubenwardy |
putting strange URLs into the address bar? Haxor |
| 20:02 |
erle |
you can convert small files to data URIs using this tool i wrote http://news.dieweltistgarnichtso.net/bin/file2datauri |
| 20:02 |
MTDiscord |
<MNH48> oh, that actually work... just put in the address bar |
| 20:02 |
MTDiscord |
<MNH48> but yes, that's correct |
| 20:03 |
erle |
https://developer.mozilla.org/en-US/docs/Web/HTTP/Basics_of_HTTP/Data_URLs |
| 20:03 |
erle |
data:[<mediatype>][;base64],<data> |
| 20:04 |
erle |
data:text/html,<script>alert('lol%20haxx');</script> |
| 20:15 |
Pexin |
I've been haxd? ._. |
| 20:31 |
|
sobkas joined #minetest |
| 20:40 |
|
sobkas joined #minetest |
| 21:19 |
|
sobkas joined #minetest |
| 21:33 |
muurkha |
erle: yes |
| 21:33 |
|
cation joined #minetest |
| 21:35 |
muurkha |
rubenwardy: Arabic speakers total about 620 million according to WP, which would be about 8% of the world, so 4.8% of the internet sounds pretty plausible |
| 21:42 |
muurkha |
https://en.wikipedia.org/wiki/List_of_languages_by_total_number_of_speakers says Ethnologue puts MSA a little lower at only 274 million (3.5%), #6 in the world, behind French (274.1 million, 3.5%), Spanish (548M, 7%), Hindi (602M, 8%), Mandarin (1120M, 14%), and of course English (1460M, 19%) |
| 21:42 |
ShadowBot |
https://github.com/minetest/minetest/issues/6 -- Apples on the trees can not be eaten |
| 21:53 |
|
sobkas joined #minetest |
| 22:00 |
|
Sven_vB joined #minetest |
| 22:01 |
MTDiscord |
<MNH48> don't forget that Arabic script issue would not just affect Arabic language, because there's a lot of languages that also use Arabic script such as Urdu, Persian, etc |
| 22:03 |
MTDiscord |
<MNH48> also the RTL issue also affect Hebrew (iirc the issue that got opened and now kept as some sort of 'tracker' was about Hebrew) so it's not just Arabic |
| 22:15 |
muurkha |
yeah, and Urdu itself is almost as big as MSA (Pakistan is huge), but Hebrew is pretty small, only about 9 million (0.1%) |
| 22:16 |
muurkha |
erle: nice to see you again |
| 22:20 |
muurkha |
luatic: I remembered today that I did once have a bug due to Lua's implicit floating point that took me an hour or two to track down. I'd hacked together a quick 2-D physics simulation: https://gitlab.com/kragen/bubbleos/blob/master/spikes/circles.lua |
| 22:20 |
muurkha |
it says it was only an hour actually |
| 22:21 |
muurkha |
the balls were bouncing in the desired way in terms of their Y coordinate, but their X coordinates were all over the place, and I was having a hard time figuring out why |
| 22:24 |
erle |
muurkha can you tell me why with the glitch music people wanted arbitrary formulas? |
| 22:24 |
muurkha |
finally I figured it out. the calculated X coordinates were fine, it was the displayed ones that were totally wrong. and they were wrong because the code to index into the pixel buffer is basically ypic.p + y * ypic.stride + x |
| 22:24 |
muurkha |
here ypic.p is actually a memory pointer handled by LuaJIT's FFI |
| 22:26 |
muurkha |
if you pass in a fractional y here you get totally bogus results; in fact if your ypic is supposed to be a rectangular window inside a larger buffer, the calculated pixel offset can actually be outside that window |
| 22:26 |
muurkha |
without so much implicit coercion it would have been easy to figure out |
| 22:26 |
erle |
muurkha have you made minetest mods? |
| 22:26 |
muurkha |
erle: what do you mean about the formulas? |
| 22:27 |
|
sobkas joined #minetest |
| 22:27 |
erle |
muurkha libglitch has no loops and the computation for the sample will always terminate in a finite number of steps (unless i dun goofed) |
| 22:27 |
muurkha |
I made a simple mod. it adds a teapot. then my girlfriend added recipes to use the teapot to make different kinds of tea |
| 22:27 |
muurkha |
erle: that's true |
| 22:28 |
muurkha |
but that's also true of arbitrary formulas, isn't it? |
| 22:28 |
erle |
muurkha check out my e guitar. simple enough to run on a microcontroller (the problem is the output though) http://news.dieweltistgarnichtso.net/bin/pluck.c |
| 22:28 |
muurkha |
dividing by 3 is bad on a microcontroller |
| 22:29 |
erle |
well, IBNIZ sucks for live music coding |
| 22:29 |
erle |
because you can get no sample |
| 22:29 |
muurkha |
so is dividing by 172 |
| 22:30 |
erle |
maybe you can fiddle with the constants a bit to make it more palatable |
| 22:30 |
muurkha |
dividing by 29412 is worse |
| 22:31 |
muurkha |
I did the same naughty division thing in http://canonical.org/~kragen/sw/dev3/ks1.c though |
| 22:32 |
muurkha |
and in http://canonical.org/~kragen/sw/dev3/ks.c |
| 22:33 |
muurkha |
and in http://canonical.org/~kragen/sw/dev3/ks-string.c |
| 22:33 |
erle |
hey yours is so much shorter lol |
| 22:33 |
muurkha |
yours might sound better though, I haven't tried it |
| 22:33 |
|
panwolfram joined #minetest |
| 22:34 |
erle |
try it |
| 22:34 |
erle |
mine is hackery upon hackery |
| 22:34 |
|
Pexin joined #minetest |
| 22:34 |
muurkha |
a better approach for a microcontroller is in http://canonical.org/~kragen/sw/dev3/ks-tiny.c |
| 22:34 |
erle |
low-pass filter? why, if i can use the % operator |
| 22:34 |
erle |
no one is going to notice anyway |
| 22:35 |
erle |
i forgot what the thing before “almost obvious” does though |
| 22:35 |
muurkha |
on an AVR the slowdown for using the % or / operator is very very noticeable |
| 22:35 |
muurkha |
especially in more than 8 bits |
| 22:36 |
muurkha |
the low-pass filter in ks-string.c was to try to get a more piano-like sound. it was not very successful |
| 22:37 |
muurkha |
yours plays an actual score, which is a lot more than any of those versions did |
| 22:38 |
muurkha |
do you know about the mini-language interpreted by the old GW-BASIC command PLAY? |
| 22:39 |
muurkha |
what were you saying about loops and arbitrary formulas? |
| 22:44 |
muurkha |
I think the |0x17 limits the range of the random number a little |
| 22:50 |
muurkha |
maybe to avoid a division by zero |
| 22:52 |
muurkha |
because at one point you assign R, the random number, to r_size, the length of the digital delay line |
| 22:53 |
muurkha |
and in several places you, naturally enough, do something % r_size to calculate a sample index |
| 22:55 |
muurkha |
which will usually result in dying with a division by zero if r_size happens to be 0 |
| 22:59 |
muurkha |
also you set damper to R periodically. not sure if anything particularly bad happens if damper - 37 goes negative |
| 23:01 |
muurkha |
by making R always return a smallish value, I did get it to start generating white noise instead of bass guitar sounds |
| 23:06 |
|
sobkas joined #minetest |
| 23:12 |
muurkha |
a general strategy for getting a digital delay line synthesizer with variable delay without division is to use a power-of-2 buffer size and calculate the lagged sample as buf[t - lag & bufsiz - 1] |
| 23:13 |
|
AliasStillTaken joined #minetest |
| 23:19 |
|
wsor joined #minetest |
| 23:19 |
|
Edgy2 joined #minetest |
| 23:19 |
muurkha |
oh, the white noise is because it's fine for damper - 37 to be negative occasionally but if damper - 37 is always negative it resets every sample |
| 23:41 |
|
sobkas joined #minetest |
| 23:48 |
|
Lesha_Vel joined #minetest |
| 23:48 |
|
specing_ joined #minetest |
| 23:49 |
|
Lesha_Vel joined #minetest |
| 23:51 |
|
Lesha_Vel joined #minetest |
