Brassica SCA [v1.0.0]

[Lērisama]

I tried that and got the following error:

Code: Select all

cli-pb/Main.hs:40:13 error:
    • Variable not in scope
        asum :: [Parser ParadigmInput] -> Parser ParadigmInput
    • Perhaps you meant one of these:
        'sum' (imported from prelude), 'asumC' (imported from Conduit)
      Perhaps you want to add 'asum' to the import list in the import of
      'Data.Foldable' (cli-pb/Main.hs:8.1-29).
   |
40 |         <$> asum
   |             ^^^^

And then error messages caused by the above.
This is weird, as it seems to be an error in the compiling, which I'm sure worked fine when I tried it yesterday; I'm not sure what's going on.

[bradrn]

This is weird, as it seems to be an error in the compiling, which I'm sure worked fine when I tried it yesterday; I'm not sure what's going on.

This is in the paradigm builder (in directory cli-pb). Perhaps you didn’t try building it yesterday.

It’s a strange error though… basically it’s saying that it can’t find the function asum, which is weird since it’s a common function. My best guess is that you’re somehow stuck on a very old (pre-4.16) version of the base library, which didn’t define asum. What does ghc --version give you?

[Travis B.]

One thing I should recommend is downloading GHC etc. directly rather than using what comes with your distros -- distros often lag behind, especially Debian-based ones (cough cough "stable" cough cough).

[Lērisama]

It claims to be The Glorious Glasgow Haskell Compilation System, version 9.0.2

[bradrn]

One thing I should recommend is downloading GHC etc. directly rather than using what comes with your distros -- distros often lag behind, especially Debian-based ones (cough cough "stable" cough cough).

Wait… this makes me realise… I thought I recommended GHCup already. But looks like I didn’t — no wonder Lērisama is having problems with painfully old GHC versions.

Lērisama, if you’re not already using GHCup, clean out your existing Haskell installation completely and reinstall it via GHCup. It should solve your problems. Distro packages of Haskell are famously bad.

[Lērisama]

Wait… this makes me realise… I thought I recommended GHCup already. But looks like I didn’t — no wonder Lērisama is having problems with painfully old GHC versions.

Lērisama, if you’re not already using GHCup, clean out your existing Haskell installation completely and reinstall it via GHCup. It should solve your problems. Distro packages of Haskell are famously bad.

Aha, I was not aware of that. I'll try that at the weekend.

[Lērisama]

I managed to install ghcup tonight, and tried to recompile it. I got a lot of

Code: Select all

/usr/bin/ld.gold: error: cannot find -lgmp
collect2: error: ld returned 1 exit status
`gcc' failed in phase `Linker'. (Exit code: 1)

As well as warnings that ~ would soon cause an error, that memset ans memcpy is deprecated. It is also complaining that importing Data.Semigroup is redundant.

[Travis B.]

I managed to install ghcup tonight, and tried to recompile it. I got a lot of

Code: Select all

/usr/bin/ld.gold: error: cannot find -lgmp
collect2: error: ld returned 1 exit status
`gcc' failed in phase `Linker'. (Exit code: 1)

As well as warnings that ~ would soon cause an error, that memset ans memcpy is deprecated. It is also complaining that importing Data.Semigroup is redundant.

You probably need libgmp installed then.

[Lērisama]

I managed to install ghcup tonight, and tried to recompile it. I got a lot of

Code: Select all

/usr/bin/ld.gold: error: cannot find -lgmp
collect2: error: ld returned 1 exit status
`gcc' failed in phase `Linker'. (Exit code: 1)

As well as warnings that ~ would soon cause an error, that memset ans memcpy is deprecated. It is also complaining that importing Data.Semigroup is redundant.

You probably need libgmp installed then.

I had just shut the pi down. Sod's law strikes again!

[Lērisama]

It works!

[Raphael]

It works!

*manic laughter*

[bradrn]

It works!

Amazing!

[Lērisama]

I have two small feature requests today. The first is something that is actually impossible to do in Brassica currently, although it is rare that the characters affected would be used after the words are in a SCA-friendly format of any kind, but the second is just a very mild shortcut

• It would be useful to be able to refer to the literal values of the characters brassica reserves, especially as part of brassicating the input, so I can write rules such as -x "*" /¹ or -x "-" / # clitic #
• It would also be useful to be able to use {…} outside of an inline category, specifically for use with the Kleene star, so an expression such as {(C) -str (C)}* would evaluate to ‘any number of of unstressed syllables, including 0’², which I currently write as [{(C) -str (C)}]*, with an annoying extra set of square brackets

¹ For when you stupidly included an asterisk at the start of every dictionary entry for an in-world reconstucted language or alternatively if your working from an existing dictionary for an actually reconstructed language
² Assuming maximal CVC syllable structure

[bradrn]

It would be useful to be able to refer to the literal values of the characters brassica reserves, especially as part of brassicating the input, so I can write rules such as -x "*" /¹ or -x "-" / # clitic #

Fair enough. For the moment I can suggest find-and-replace. (Or sed, if you like).

For syntax, I’m not sure I love double-quotes. What would you think of using an escape character, say \* or $*?

It would also be useful to be able to use {…} outside of an inline category, specifically for use with the Kleene star, so an expression such as {(C) -str (C)}* would evaluate to ‘any number of of unstressed syllables, including 0’², which I currently write as [{(C) -str (C)}]*, with an annoying extra set of square brackets

Shouldn’t be too hard to implement… though first I’d like to satisfy myself that there’s no unexpected interactions with anything else.

(BTW, I haven’t been working on Brassica at all recently — I just haven’t had the time. But these at least should be fairly straightforward.)

[Lērisama]

Fair enough. For the moment I can suggest find-and-replace. (Or sed, if you like).

I do use it already, it's just a pain having to do the copy-pasting (which isn't an issue for sed, but I don't always use the command line version). It's not at all urgent, just a way to lessen some tedium.

For syntax, I’m not sure I love double-quotes. What would you think of using an escape character, say \* or $*?

I don't like them either; I considered using an escape character in my examples, but couldn't think of a good one that wasn't taken. \* is probably the best, as $ has a use already. I only rejected it initially because it's a bit mistakeable for the very important /, but if it's always appearing before other characters, that's shouldn't be a problem.

It would also be useful to be able to use {…} outside of an inline category, specifically for use with the Kleene star, so an expression such as {(C) -str (C)}* would evaluate to ‘any number of of unstressed syllables, including 0’², which I currently write as [{(C) -str (C)}]*, with an annoying extra set of square brackets

Shouldn’t be too hard to implement… though first I’d like to satisfy myself that there’s no unexpected interactions with anything else.

Thank you

(BTW, I haven’t been working on Brassica at all recently — I just haven’t had the time. But these at least should be fairly straightforward.)

I hope your life has more free time in the future.

[Neonnaut]

For syntax, I’m not sure I love double-quotes. What would you think of using an escape character, say \* or $*?

Perhaps the backtick.

[bradrn]

For syntax, I’m not sure I love double-quotes. What would you think of using an escape character, say \* or $*?

Perhaps the backtick.

Perhaps! Except that I worry that the backtick itself might occur in words reasonably often (e.g. for tones)…

[Lērisama]

For syntax, I’m not sure I love double-quotes. What would you think of using an escape character, say \* or $*?

Perhaps the backtick.

Perhaps! Except that I worry that the backtick itself might occur in words reasonably often (e.g. for tones)…

The obvious solution would be \\, but I don't particularly like that.

[Lērisama]

What about using & as the escape character? It already has a similar but not confusible use in brasica, to in effect force featural categories to be read as nom-featural ones¹, but you can't use a category as a grapheme, and the context is different – inside, instead of (mostly) outside inline categories, and it's only use in phonetic transcription I'm aware of is as the X-SAMPA for the so-rare-its-existence-is-disputed¹ ⟨ɶ⟩, so it's unlikely it's been used in anyone's sound changes.

¹ Yes I know none of the vowels exist in the way the IPA would like to believe, but I don't think there is even an acknowledged range of F1 & F2 values to label ⟨ɶ⟩

[Man in Space]

I've discovered Brassica can accept some outré characters—for my latest work I've successfully employed ⟨∏⟩ as a category identifier.