1. Re: Kat and sequence ie EU vs Perl
Kat wrote:
>I didn't mention gamers, but the same hold true for them too. Ai engines are
>used in some games online to give characters character. 
But
>specifically, i mentioned strings in NLP. David is still correct in saying
>Eu sequences are powerful in Ai, i agree, especially in data storage, tags,
>etc where you need to be able to add/tag unknown data to unknown data at
>runtime. Testing sequences using bytes or even bits is useful too, so the
>lack of a type is handy,, i got so tired of running into type errors.
>Sequences can hold any data record, and used in whatever way you want to.
>Now if Eu can grow to use sequences and sequence names *generated* at
>runtime, when the programmer didn't know it was needed at compile time....
>See code at the bottom that i used tonite.
snip
>The way i see it, *unrestricted* strings are one step above sequences, even
>if they stay sequences, and we only *handle* them as strings. See how a few
>choice words makes it easier to parse info out of a series of bytes?
>
>Makes coding so much easier when your mind is on the words and how they
>relate to each other, and not the index of the start of the possible word in
>a sequence of bytes.
Very well put.
>PS, i seldom do the following, i was demo'ing something, but it shows
>building a var name that i wouldn't know at run time:
snip
>/ /if ( == ) { %testcommand }
>
>and the string %testcommand is executed.
>But that's not Eu. Yet?
>/me looks wishful.
>
Everybody please don't pile on this, because REXX has it's own problems,
but REXX implements both of the above capabilities. Eu does not appear
to parse a line but once, making it in some ways more a compile and go
or incremental compiler than an interpreter in final effect. True value
orientation allows your percent prefixed variable to occur "anywhere" to be
evaluated and then re-evaluated as often as necessary to eliminate
evaluatable expressions and then executed. REXX allows the programmer to
limit the depth of this process to stop infinite evaluation loops. It also has
error routines for recovery from the situation where overflow is reached.
The most interesting statement in REXX is the parse command that allows
very powerful and compact tokenizing of text. The insitu processing that
you described in your post is in some ways even more useful.
I'm hoping that the routines that you refer to above are amongst those that
you received and will be posted to the archive.
Everett L.(Rett) Williams
rett at gvtc.com
