Just a thought, would it be wise to add a warning if a
Library has more than one Path in logtalk_library_path/2? Loading order matters; only the first one would be loaded, and it probably indicates a Library name clash. Also, I see packs are working with a logtalk_packs library path, would that be changeable in settings.lgt so I could install packs on a per project basis? If so, that's going to be a really nice way to handle both global and local package management.
😎
(Just a little joke at your expense, @PaulBrownMagic: @pmoura has a tendency to already have features people request. :D)
When you're not exploring the available resources but looking for something specific, there is https://logtalk.org/search.html
Not sure what you mean by bad syntax. You example is interesting:
$ gprolog
GNU Prolog 1.5.0 (64 bits)
Compiled May 8 2021, 10:47:25 with gcc
Copyright (C) 1999-2021 Daniel Diaz
| ?- [user].
compiling user for byte code...
;- ima(Joe,K).
user:1:2: syntax error: previous operator needs brackets
But:
$ swipl
Welcome to SWI-Prolog (threaded, 64 bits, version 8.3.24-30-gaecdf37df)
SWI-Prolog comes with ABSOLUTELY NO WARRANTY. This is free software.
Please run ?- license. for legal details.
For online help and background, visit https://www.swi-prolog.org
For built-in help, use ?- help(Topic). or ?- apropos(Word).
?- [user].
|: ;- ima(Joe,K).
Warning: user://1:8:
Warning: Singleton variables: [Joe,K]
|: ^D% user://1 compiled 0.01 sec, 1 clauses
true.
Permissive systems are the problem in this particular case.
(Pronounce ima(Joe,K) out loud.)
Wait, what? That's valid syntax in SWI!? :-o
Now I can't even guess what it's supposed to mean.
Semicolon hyphen term?
Even weirder. SWI is inconsistent.
?- ;- ima(Joe,K).
ERROR: Unknown procedure: (-)/2 (DWIM could not correct goal)?- ; true.
ERROR: Syntax error: Operator expected
ERROR: ;
ERROR: ** here **
ERROR: true .
Which it is in other contexts. This is getting very much into the realm of rabbit holes.
I've switched from swilgt to ciaolgt and I'm encountering a trouble with phrase/3 and phrase/2 :
?- phrase(langclex::string(T), "toto").
{ERROR: user:::/4 - existence error: procedure:user:::/4 does not exist}
aborted:-/
The reason that goal works with
swilgt is because the SWI-Prolog adapter file contains goal-expansion code to deal with it.
A portable top-level workaround isto use instead
logtalk << phrase(…).
But the best solution is usually to define a public predicate in the object that calls the
phrase/2-3 predicates.
See e.g.
examples/dcgs in the Logtalk distribution for some examples.
I'm mostly interested in what i can (reasonably) assume is portable. Is there a good place to reference the ISO standard? I also saw that the recent release notes mention that
between/3 is a de facto standard. Is there anything outside of good-old experience that would help me pick out these de facto standards?
The
logtalk3/tests/prolog includes tests for both ISO and de facto standard features. If there's a test set for the feature you want to use, you can assume some portability at least.
Otherwise, you can always ask.
I'm looking now and it seem like
ensure_loaded/1 and acyclic_term/1 are the only two things that come up in a search for "not supported"
Hello, there is quick start for supported engine, is there something for unsupported engine?
It's unknown. I wanted to how is a port done to an unknown unsupported backend.
A little gif of where I'm at with the frames work: https://twitter.com/PaulBrownMagic/status/1401308039131668487?s=20