[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [libreplanet-discuss] Fwd: Libre Business for the Planet

From: Thomas HARDING
Subject: Re: [libreplanet-discuss] Fwd: Libre Business for the Planet
Date: Wed, 11 Mar 2015 23:25:39 +0100
User-agent: Mozilla/5.0 (X11; Linux i686; rv:31.0) Gecko/20100101 Icedove/31.5.0

On 11/03/2015 18:09, Logan Streondj wrote:

about SPEL:
here is the main project site:

On Tue, Mar 10, 2015 at 07:59:20PM +0100, Thomas HARDING wrote:
[with my tedious English]
> Where the point is missed is : at low level (effective: excerpt for
> libraries function names), most of languages has for name entries
> a very limited set of "keywords". Much more : even translated, these
> keywords will not follows in any way orthographic nor syntactic
> natural language rules (plural, gender, terses, ...).

SPEL currently already supports 25+ linguistic universals.
Though for simplicity the input language is analytic,
where the grammar words are seperated from the vocabulary words,
such as in Japanese, Chinese, Hebrew and to lesser extent English.

One good point to that project is : it follows native character sets (probably text direction too).

*In fact, the best language fitting your goal would be an "Espéranto" derivative.*

As far as I having make programs, current advantage I give to Englsh is : each language keyword is
"enclosed" by an unambiguous sense. Programmatically unambiguous.

Moreover, in my attempts to translate some RFC's, some terms are defined unambiguously by long definitions (MUST, MUST NOT, MAY...). So unambigueously that preliminary definitions
has now their own RFC document.

i give such French sample sentences later, as your samples souds like (at least in French) Pidgin (to a French they just look likes a complete mess). To produces that by hand I'd need probably
one of the numerous automatic translators we can found aroun the Ternet.

Russian has for grammar basis declension.

Also the case in German, then needs most of times to complete the latest verb of sentence in order to not be
misunderstood (little soldier story).

My concerns are : (key)words disambiguation, non-natural expression (just think any sentence won't "compile" the same in any human brain, as far as i need to type a <g> here just in case you'll read the whole as a rant
--- or worse, a joke).

Also, there is an expression named "franponais", where some Japanese poetry is translated programmatically to French for use as hostels, restaurants... Unfortunately there are no more means in while typed in French.

> I taked a look at code from a colleague written in a French proprietary
> ide/language named "Windev". Keywords are French, but it (sintactically)
> have nothing to do with French language.

So let me give 2 examples,
a hello world, and a simple sentence:

be say ob tha hello world ya

"be" indicates the verb phrase,
"ob" indicates the grammatical-object,
"tha" is the kind of "that" which starts a subordinate clause,
"ya" is sentence-final-particle analagous to a period.

That is definitely non-natural language.
here it is in the 6 UN langauges,
plus mwak which is the pivot language:

eng : be say ob tha hello world ya
zho : 世界 您好 之 对象 说 是 啊
spa : ser decir ob que hola mundo sí
ara : يكون قول أوب ثا مرحبا عالم يا
rus : быть сказать о то привет мир да

fra : être dire ob que bonjour monde ya
A human comprehensive French sentence would be :

 "Cet objet dit : « Bonjour au monde entier »."


"Ceci dit « bonjour au monde entier »" (That says: "Hello World").

 or precisely:

On inscrit par cette définition « Bonjour au monde entier » ; dans le cas où le périphérique de sortie serait un synthétiseur de paroles, la déclaration « inscrit » sera remplacée par « prononce ».

Formally and complete :

L'objet déclaré dans cette phrase a pour fonction de faire apparaître les mots "Hello World", à un endroit et par des moyens non encore définis.

mwak : munt sla ti .a yan .i ya

It can compile to javascript as
say(gettext("hello world"));

now for an example of use in conversation:

su me be go to the shop for bread ya

"su" is grammatical subject.

this is the input-version in the UN languages:

eng : su me be go to the shop for bread ya
zho : 面包 为 我 主题 去 是 店 该 到 啊
spa : su me ser ir a la tienda para pan sí
ara : يكون اذهب سو أنا إلى ال متجر سبيل خبز يا
rus : а мне быть идти к эт магазин для хлеб да
fra : su moi être aller à la boutique pour pain ya
su(jet) : Je me suis rendu à la boutique afin d'acheter du pain.

Curiously, I authored in Bash for training-course purpose a "acheterpain"/"bybread" command, with an extensive set of options/arguments/, producing mail, message (msg y), ... and sms (thanks to gsmutils gsmsendms).

Goal was to make students understands the "verb complements subjects" general syntax of shell commands, moreover to produce human comprehensive texts depending on gender, age, with for fallback ...the house dog;

mwak : panp plu mi .u tsuk sa ta kiy .i ya

currently there is also primitive conjugation support for output,
makes it look a little more natural:

eng :  I be go to the shop for bread.
zho : 面包 为 我 去 是 店 该 到 啊
spa :  soy ser ir a la tienda para pan.
ara : يكون اذهب أنا إلى ال متجر سبيل خبز.
rus :  я быть идти к эт магазин для хлеб.
fra :  je être aller à la boutique pour pain.
mwak : panp plu mi .u tsuk sa ta kiy .i ya

conjugation is currently translation-memory bases,
so basically a simple search and replace (su me becomes I).

better conjugation will be added after compiling down to programming
languages works at a functional level.

What must compile nicely..

> On the other hand, I heard for era about "cweb", from D. Knuth,
> (but never took a look at any cweb code, I'm lazy...). The described
> approach is ... to describe what does code does as long as writing
> the program itself, then pre-process that we call otherwise "comments"
> and nested code to produce the code itself, then compile.

CWEB is literate programming, which is a fancy name for saying it is
easier to write comments than code. Makes it more like writing a
scientific paper, with a few formulas.

In other words, a manner to "SAY" constantly what it will "DO".

> Unfortunately I found only references on Java/Spring about "SPEL".
> Filtering out gives for result "SIGPLAN" and Spel workgroup.
> If not based on coffee, I'm curious of a bunch of URI you could post
> Regards,

[URIs rewritten at begin]
from Logan ya

reply via email to

[Prev in Thread] Current Thread [Next in Thread]