From Newsgroup: comp.lang.prolog

Hi,

What makes the Prolog community probably so
dumbed down, concerning parsing and language theory.
Is possibly the blunder of focusing on

rCLconstraint programmingrCY in the 80s and 90s.
So the Gen-X and later generations has become
clueless about parsing and language theory.

Alain Colmerauer lists in his China talk:

- PrologIA
- ILOG
- COSYTEC

But maybe dumbos like Julio Di Egidio
<julio@diegidio.name> are the exception? The
problem roots deeper. Not only has the Prolog

community become dumbed down, about to
use DCGs in production, it also doesn't trust
a single Prolog system to do it efficiently

to have it used in production, like the realization
of a Prolog system itself. Somehow methods and
tools to realize efficient DCGs in Prolog are

missing. Most DCG attempts that one sees succumb
to some declarative nonsense, creating exponentially
many spurious choice points, you find rarely

somebody mastering the Art.

Bye

Mild Shock schrieb:

Hi,

I didn't unlock the Metamorphoses Paper yet,
its still behind a paywall for me. But
a Paper about W-Grammars is available.

The Paper about W-Grammars is from
Alain Colmerauers website again.
While a W-Grammar rule of the form:

X |-> Y

Leads mathematically to a rewrite rule:

V,X,W ==> V,Y,W

A DCG Transform of the form:

P,Y --> X,P

Leads mathematically to a rewrite rule:

V,*,X,W ==> V,Y,*,W

Where * is a kind of replacement cursor.
So in DCG Transform there is no comming back
to match and replace a previous position again.

Thats why in my NetFish Transducer, I need
"cascading", basically repeatedly applying
the DCG Transforms by iteration,

until a kind of fixpoint is reached.

Bye

See also:

W-Grammar, Chastellier & Colmerauer (1968) http://alain.colmerauer.free.fr/alcol/ArchivesPublications/Wgrammar/Wgrammar.pdf

Mild Shock schrieb:

Hi,

I only rediscovered a little bit that DCG has
also semi-context, and that when using non-terminals
in semi-contexts, these act like what

was Meta-Grammars in W-Grammar, 1967rCo1968.
W-Grammars with Meta-Grammars were basically
Semi-Thue systems, but one needs to pay attention,

the todays DCG cannot directly implement them:

-aFrom natural language processing to Prolog
http://alain.colmerauer.free.fr/alcol/ArchivesTransparents/ChinaApril2011/NatLanguageProlog.pdf


What later came was Q-System, 1968rCo1970, French,
1971rCo1973, Prolog, 1973rCo1976 and Prolog II,
1977rCo1982. An early shift to CLP(X) was dif/2

which already appeared in Prolog II. That todays
Prolog commmunity is clueless about the grammar
and transform aspect is really sad.

Bye

Mild Shock schrieb:

Hi,

Only morons waste time with regex when they have Prolog!
Its amazing, since Alain Colmerauer was already
juggling with a kind of transducers:

"We present some very general grammars in which
each re-writing rule is of the type: replace such
and such sequence of trees by such and such another
sequence of trees."
https://www.researchgate.net/publication/225124810

Still the Prolog community is to stupid to
transduce CR, CR LF and LF into LF. Thats quite a
feat. BTW, this should do using DCG semi-contexts:

replace, [0'\n] --> [0'\r, 0'\n], replace.
replace, [0'\n] --> [0'\r], replace.
replace, [0'\n] --> [0'\n], replace.
replace, [X] --> [X], replace.
replace --> [].

Works on my side:

?- replace("foo\rbar\r\nbaz", _L), atom_codes(A, _L).
A = 'foo\nbar\nbaz'

Now optimize it to produce 0'\n as early as
possible and use a state machine, so that you
can implement it low level in your streams.

Have Fun!

Bye

--- Synchronet 3.21a-Linux NewsLink 1.2