The Computer And General Tech Thread - Software, Hardware, Questions, etc.

[Vilike]

I would be impressed with an SCA with real stress and suprasegmental support at all, since this is an area which SCA's are notoriously poor at, and which is why I don't use an SCA for Laqar despite it seeming like it would benefit from one with the complexity of its diachronics.

Lexurgy SC. The ability to define a feature as floating (thus not affected by a majority of sound changes), the filter rules forgoing the need to write optional segments if they're not affecting the change (like consonants in tone sandhi), and the "propagate" option to apply

For example, some simple stress assignment rules:

Code: Select all

#Distinctive features on segments (* means default feature if unmarked)
Feature Type(*cons, vowel)
Feature Stress(*unstressed, secondary, primary)

# Defining segments for testing an example
Symbol a [vowel]
Symbol t []

Diacritic ˈ (floating) [primary]
Diacritic ˌ (floating) [secondary]

primary-stress-second-last-syllable [vowel]:
# Applies antepenultimate primary stress in polysyllables and final in monosyllables
 [] => [primary] / _ [] $
 [] => [primary] / $ _ $

add-secondary-stress [vowel] propagate:
 # Applies secondary stress to alternating vowels before the main stress.
 [unstressed] => [secondary] / _ [] {[primary], [secondary]}
 

Testing it on:

Code: Select all

tat
ta
tata
tatatatatatata

gets you:

Code: Select all

tat            => taˈt
ta             => taˈ
tata           => taˈta
tatatatatatata => tataˌtataˌtataˈta

(Okay, how it's coded makes the diacritic follow the stressed vowel, unlike in IPA where it precedes the syllable. But if I'd use acute and grave accents, that would've been prettier)

Give it a try!

[Travis B.]

Question - how do you write it such that stress influences segments not adjacent to the segment the diacritic is on, e.g. aspirating a fortis explosive separated from the vowel by other members of a consonant cluster?

[Travis B.]

I have gotten the block interface mostly working, as far as I can tell (even though to be sure I need to do more work). Note that it does not really implement a linear array of 1K blocks but rather a map of values from 0x00000000 to 0xFFFFFFFE (the main change I know I need to make so far is to forbid the user from specifying 0xFFFFFFFF). There is no need for values to be contiguous. One cannot use anywhere near all the possible values at once because there will not be enough space for all them.

[Zju]

On the topic of SCAs, are there any that are running purely on JS? Lexurgy seems to be transpiled to JS, so that's one, even if it was written in Kotlin. At a quick glance, SCA² appears to be JS only. Any others?

[Vilike]

Your Conlangjugator, of course. And that's all I could find.

[Travis B.]

zeptoforth 0.14.1 is out! It adds Quad SPI flash and block support for the STM32F746 DISCOVERY board, and fixes an issue which resulted in broken non-kernel-only binaries for the STM32F746.

[Richard W]

On the topic of SCAs, are there any that are running purely on JS? Lexurgy seems to be transpiled to JS, so that's one, even if it was written in Kotlin. At a quick glance, SCA² appears to be JS only. Any others?

Well, there's the translation that I put up over 17 years ago. Unfortunately, I couldn't see any way to get it to work with files in a civilised fashion.

[Travis B.]

I was originally going to improve the line editing capabilities of the zeptoforth kernel (in ARM Thumb-2 assembly) so it would properly handle deleting tabs, but as I couldn't make that work I switched to a different approach, which is to write a much more capable line editor in Forth, which is a cut-down version of a line editor I wrote for a different Forth, that uses the REFILL hook to replace the very minimal zeptoforth kernel line editor. At the moment it does not really work, but at the same time it is more promising than what I had been trying to do.

[Travis B.]

zeptoforth 0.14.2 is out. It now includes kitchen-sink "big" binaries, a block interface (for the STM32F746) using Quad SPI, a basic line editor, task pools, and a fix to CASE/OF/ENDOF/ENDCASE.

[Travis B.]

zeptoforth 0.14.3 is out! It now includes a block editor for the STM32F746 board.

[Travis B.]

I now have support for escaped strings (through using .\" , s\" , c\" , and .\( ) in zeptoforth, allowing things such as multiline strings, strings containing double-quotes, and strings containing special characters, which will be available in all of the builds once I make a new release. I will also, prior to making a release, include more works for manipulating blocks (e.g. copying blocks, moving blocks, inserting blocks, etc.)

[Travis B.]

zeptoforth 0.15.0 is out! The primary change here is the addition of a new module system to zeptoforth, which the codebase now uses as a whole (except for basic.fs, because it is loaded before the module system).

[Travis B.]

And zeptoforth 0.15.1 is already on its way, since I decided to make a significant identifier naming change globally just after releasing zeptoforth 0.15.0...

[Travis B.]

And zeptoforth 0.15.1 is out, with renamed module identifiers (from *-wordlist to *-module).

[Travis B.]

zeptoforth 0.15.2 is out, now including action pools, a GPIO API, adjustable timeslices for tasks, support for more efficient use of external flash memory when available space is tight, the ability to close (and reopen) channels and fast channels, and support for inserting and deleting rows in the block editor.

[Travis B.]

I have a working true random number generator implementation working for the STM32F407 for zeptoforth, and it should not be hard to port it to the STM32L476 and the STM32F746.

[bradrn]

I have a working true random number generator implementation working for the STM32F407 for zeptoforth, and it should not be hard to port it to the STM32L476 and the STM32F746.

‘True’? In that case, what’s the source of randomness?

[Travis B.]

I have a working true random number generator implementation working for the STM32F407 for zeptoforth, and it should not be hard to port it to the STM32L476 and the STM32F746.

‘True’? In that case, what’s the source of randomness?

It's the hardware itself. What I have implemented is a mere driver that accesses hardware registers to get random numbers.

[bradrn]

I have a working true random number generator implementation working for the STM32F407 for zeptoforth, and it should not be hard to port it to the STM32L476 and the STM32F746.

‘True’? In that case, what’s the source of randomness?

It's the hardware itself. What I have implemented is a mere driver that accesses hardware registers to get random numbers.

Sure, I understand that. What I meant to ask was what method was used within the hardware to generate entropy (e.g. Cloudflare famously uses lava lamps).

…except I just had the bright idea of looking up the STM32F407, and after a couple of Google searches I find the source is a complicated-looking circuit of ring modulators which I don’t understand at all, so there was no need to ask in the first place. Ah well, never mind.

[Travis B.]

Sure, I understand that. What I meant to ask was what method was used within the hardware to generate entropy (e.g. Cloudflare famously uses lava lamps).

…except I just had the bright idea of looking up the STM32F407, and after a couple of Google searches I find the source is a complicated-looking circuit of ring modulators which I don’t understand at all, so there was no need to ask in the first place. Ah well, never mind.

Yeah, I honestly do not understand how the hardware works here, hence my answer treating the hardware as a magical oracle of randomness.