p-lang thread: (now (have you (problems two)))

The Something Awful Forums > Discussion > Serious Hardware/Software Crap > YOSPOS > p-lang thread: (now (have you (problems two)))

«‹›1784 »

Destroyenator: Dec 27, 2004; Don't ask me lady, I live in beer

Symbolic Butt posted:

the most surprising part of prolog to me is that it's kind of annoying to do arithmetic in it, I'm still trying to figure out a way for unification to work with some simple maths

http://www.swi-prolog.org/pldoc/man?section=clpfd does everything you would expect from a constraint solving maths thing, you just have to use slightly different operators

# ? Mar 10, 2018 19:10

Adbot: ADBOT LOVES YOU

# ? Jun 1, 2024 04:21

tef: May 30, 2004; -> some l-system crap ->

Internet Janitor posted:

the idea of prolog is to describe your input and output and the language furnishes a program.

The idea of prolog is that well, it's not so much about your input and output but your relationships between data.

The idea of prolog is that you describe the constraints of your data, and it works out if those constraints are satisfied.

Well,

It's a database. You write queries, and you can give some queries names.

Kinda,

It's a parser. Prolog is about nested grammars. You don't write a program, but do write a grammar.

And Yes

You can do reversible programming, or program generation.

But you do it by specifying the rules and then using prolog as a rules engine.

quote:

in trivial cases this works. in meaningful cases it's intractable. in practice, the idea of prolog was a complete failure.

the idea that prolog had a single idea is kinda laughable given how it was invented twice for different reasons.

the french (well, marseilles) team had been writing a generic parser, and had come up with an annotated grammar. the uk (well, edinburgh) group had been working on resolution algorithms independently, and discovered that another group had already implemented it without a formal model.

there was the prolog fad in 1980 when we thought machine learning was impossible and so we had to build expert rule systems to solve every problem

quote:

prolog the actual language is kind of a neat way of thinking about some things, though.

the irony is that: the ideas of prolog have been a success, the implementations of prolog on the other hand...

i mean that SLD resolution, unification, logic variables show up in other places in various forms.

quote:

writing programs that are not restricted to a narrow domain is intrinsically, inescapably difficult. we still can't really teach humans how to compose complex programs, so it shouldn't be surprising that we have met with limited success in replacing programmers with programs. i think the best thing to aim for is tools and techniques to help people manage the complexity of software, rather than hoping to hide it or make it disappear via some sleight of hand.

the whole 'replacing humans with programs' thing is some weirdo sales pitch from someone desperate for funding

# ? Mar 10, 2018 19:26

tef: May 30, 2004; -> some l-system crap ->

gosh this whole 'the program doesn't describe a total ordering for evaluation' is weird

*writes a makefile and watches how the runtime sorts the build rules into an order for trying*

# ? Mar 10, 2018 19:28

DONT THREAD ON ME: Oct 1, 2002; by Nyc_Tattoo; Floss Finder

i am seriously considering prolog for a thing at work

# ? Mar 10, 2018 19:29

Gazpacho: Jun 18, 2004; by Fluffdaddy; Slippery Tilde

Can I suggest: don't

# ? Mar 10, 2018 19:30

Symbolic Butt: Mar 22, 2009; (_!_); Buglord

Destroyenator posted:

http://www.swi-prolog.org/pldoc/man?section=clpfd does everything you would expect from a constraint solving maths thing, you just have to use slightly different operators

holy poo poo this is exactly what I needed, thanks a lot

# ? Mar 10, 2018 19:33

Suspicious Dish: Sep 24, 2011; 2020 is the year of linux on the desktop, bro; Fun Shoe

We have prolog in our game engine. Don't ask me why

# ? Mar 10, 2018 20:21

aardvaard: Mar 4, 2013; you belong in the bog of eternal stench

Gazpacho posted:

Can I suggest: don't

just use bash instead

# ? Mar 10, 2018 21:07

pokeyman: Nov 26, 2006; That elephant ate my entire platoon.

Suspicious Dish posted:

We have prolog in our game engine. Don't ask me why

why

# ? Mar 10, 2018 22:17

Gazpacho: Jun 18, 2004; by Fluffdaddy; Slippery Tilde

for AI i bet

# ? Mar 10, 2018 22:24

Symbolic Butt: Mar 22, 2009; (_!_); Buglord

Gazpacho posted:

for AI i bet

my bet is on ~*~interactive fiction~*~

# ? Mar 10, 2018 22:31

Doom Mathematic: Sep 2, 2008

Suspicious Dish posted:

We have prolog in our game engine. Don't ask me why

Procedural generation of instances of some kind of puzzle which have to be guaranteed solvable?

# ? Mar 10, 2018 22:38

Suspicious Dish: Sep 24, 2011; 2020 is the year of linux on the desktop, bro; Fun Shoe

Symbolic Butt posted:

my bet is on ~*~interactive fiction~*~

ding ding ding

# ? Mar 10, 2018 23:09

tef: May 30, 2004; -> some l-system crap ->

i don't know what to make of this

like prolog is good because it is bad

# ? Mar 10, 2018 23:46

akadajet: Sep 14, 2003

i only had to look at prolog for like 1 week in college, then i've never seen it again.

# ? Mar 10, 2018 23:47

Suspicious Dish: Sep 24, 2011; 2020 is the year of linux on the desktop, bro; Fun Shoe

Yeah, don't ask me how it worked, it was another team that worked on that project.

code:

expression_of_displeasure(_Data, A)-->
    ["[[ ", [subject,standard]-A, " scowls disapprovingly ]]"];
    ["[[ ",[subject,standard]-A, " rolls ", [object,possessive]-A, " eyes ]]"];
    ["[[ ",[subject,standard]-A, " sighs in frustration ]]"];
    ["[[ ",[subject,standard]-A, " scowls disapprovingly ]]"];
    ["[[ ",[subject,standard]-A, " scowls disapprovingly ]]"];
    ["[[ ",[subject,standard]-A, " forces a smile ]]"];
    ["[[ ",[subject,standard]-A," casts a sad look in your direction ]]"].

# ? Mar 11, 2018 00:31

tef: May 30, 2004; -> some l-system crap ->

that's like a DCG but

i guess text generation makes some sense but wheee

# ? Mar 11, 2018 00:33

ynohtna: Feb 16, 2007; backwoods compatible; Illegal Hen

wow, prolog is hella overkill for that purpose.

# ? Mar 11, 2018 00:36

Suspicious Dish: Sep 24, 2011; 2020 is the year of linux on the desktop, bro; Fun Shoe

i assure you the actual actual prolog is a lot more complicated than that. it drove game state.

# ? Mar 11, 2018 00:46

redleader: Aug 18, 2005; Engage according to operational parameters

Glorgnole posted:

labview doesn't let you zoom because lots of its interpretation of what the blocks in the window are supposed to do comes down to the relative positions of the pixels in the icons and wires

jesus loving christ

# ? Mar 11, 2018 02:52

Presto: Nov 22, 2002; Keep calm and Harry on.

akadajet posted:

i only had to look at prolog for like 1 week in college, then i've never seen it again.

I had a whole semester with it. All I remember is writing a program to prove a guy was his own grandfather.

# ? Mar 11, 2018 04:53

eschaton: Mar 7, 2007; Don't you just hate when you wind up in a store with people who are in a socioeconomic class that is pretty obviously about two levels lower than your own?

?-

eschaton fucked around with this message at 10:21 on Mar 11, 2018

# ? Mar 11, 2018 10:19

raminasi: Jan 25, 2005; a last drink with no ice

I prototyped an algorithm in Prolog once before implementing it for real, it worked really well.

also re visual programming: it�s a fantastic gateway drug for people who do complicated enough (domain-specific) work that it calls for programming but are intimidated by or disinterested in �programming� as an activity. �oh I�m not programming, this is �parametric modeling�.� �oh I don�t know much about programming, this is just a bespoke multimedia controller.�

# ? Mar 11, 2018 14:51

pokeyman: Nov 26, 2006; That elephant ate my entire platoon.

eschaton posted:

?-

amberpos.css needs to swap that out for a correct gif

# ? Mar 11, 2018 15:03

akadajet: Sep 14, 2003

pokeyman posted:

amberpos.css needs to swap that out for a correct gif

i think amberpos being ever so slightly broken is the correct behavior

# ? Mar 11, 2018 15:08

Symbolic Butt: Mar 22, 2009; (_!_); Buglord

I've been playing with prolog lately to see if I get a deeper feel of the language, I'm doing this gameboy z80 assembler thing, that I hope eventually will result in a bespoke artisanal compiler.

so from a list of instructions I'm getting a list of assembled code.

code:

?- compile_list([instruction(ld, a, b), instruction(ld, d, 4), instruction(inc, de), instruction(inc, hl), instruction(dec, bc), instruction(nop)], X).
X = [120, 22, 4, 19, 35, 11, 0] ;

and I can do the opposite, if I did this right writing an assembler in prolog gives you a disassembler for free.

code:

?- compile_list2(X, [120, 22, 4, 19, 35, 11, 0]).
X = [instruction(ld, a, b), instruction(ld, d, 4), instruction(inc, de), instruction(inc, hl), instruction(dec, bc), instruction(nop)] ;
false.

but I messed something up and I'm trying to figure out what. I needed to write 2 separate rules for this, inverting the order of the facts

code:

compile_list(Instructions, AssembledList) :-
    encode_list(Instructions, MachineInstructions),
    assemble_list(MachineInstructions, AssembledList).

compile_list2(Instructions, AssembledList) :-
    assemble_list(MachineInstructions, AssembledList),
    encode_list(Instructions, MachineInstructions).

I'll keep the thread updated with my incompetent coding adventures

# ? Mar 13, 2018 00:58

Destroyenator: Dec 27, 2004; Don't ask me lady, I live in beer

it looks like it's all valid, you just aren't constraining the search well enough when you're going backwards. it's trying to build any list of instructions and then checking if they're a match for the assembled output you're providing

if you interrupt it and press g you can see what it's stuck trying to solve, it's just building an ever growing list of "nop"s and then testing that against the output list

code:

?- compile_list(X, [120, 22, 4, 19, 35, 11, 0]).
^CAction (h for help) ? goals
  [9] assemble_list([machine_instruction(opcode(0, 0, 0)), machine_instruction(opcode(0, 0, 0)), machine_instruction(opcode(0, 0, 0)), machine_instruction(opcode(0, 0, 0)), machine_instruction(opcode(0, 0, 0)), machine_instruction(opcode(0, 0, 0)), machine_instruction(opcode(..., ..., ...)), machine_instruction(...)|...], [120, 22, 4, 19, 35, 11, 0])
  [8] compile_list('<garbage_collected>', [120, 22, 4, 19, 35, 11, 0])

# ? Mar 13, 2018 08:55

Symbolic Butt: Mar 22, 2009; (_!_); Buglord

yeah that gives me a much better idea of what's going on now, thanks for the tip

# ? Mar 13, 2018 16:34

Gazpacho: Jun 18, 2004; by Fluffdaddy; Slippery Tilde

Notorious b.s.d. posted:

we established long ago that you are too badly damaged by your career to understand the difference between good and bad.

btw good tech is what my employer allows me to use. bad tech is what i might wish i could use but would do better to just not think about lest i go mad

# ? Mar 14, 2018 19:37

Notorious b.s.d.: Jan 25, 2003; by Reene

Gazpacho posted:

btw good tech is what my employer allows me to use. bad tech is what i might wish i could use but would do better to just not think about lest i go mad

Notorious b.s.d. posted:

we established long ago that you are too badly damaged by your career to understand the difference between good and bad.

# ? Mar 14, 2018 20:08

tef: May 30, 2004; -> some l-system crap ->

i wrote a network transparent option parser

https://github.com/tef/textfree86

:2bong:

# ? Mar 17, 2018 08:28

tef: May 30, 2004; -> some l-system crap ->

i'd post in the job thread or terrible thread but this is kinda an unspeakable horror that deserves to be here, i guess

here's the gist

if you have a command line program, you can expect some things to work: passing in a file name, and if you're lucky, tab completion:

code:

$ ./logdetails --uptime log.1
Wrote uptime to log

$ cat log.1
 5:59  up 2 days, 13:31, 12 users, load averages: 0.83 0.93 1.05

# Tell bash how to find completion information: (logdetails does it!)
$ complete -o nospace -C ./logdetails ./logdetails

$ ./logdetails --u<TAB>     # Using Tab Completion
--uname   --uptime

If you have a command line program, and you run it via ssh, or docker run, some things just don't work the same.

code:

$ ssh machine ./logdetails filename   # filename is local to that machine

$ ssh machine ./logdetails  --<TAB>   # tab completion is for ssh, not grep

If you use textfree86, well, some things do work the same:

code:

$ ./textfree86.py 'ssh hostname /path/to/logdetails --pipe' -- output.log --uname
Wrote uname to log

$ cat output.log    # Written on the local machine!
Darwin

# We add alias, so we can add bash completion
$ alias rlogdetails='./textfree86.py ssh hostname /path/to/logdetails --pipe --'

$ rlogdetails --help
usage: logdetails [--uptime] [--uname] <output>

$ complete -o nospace -C rlogdetails rlogdetails
$ rlogdetails --u<TAB>    # Using Tab Completion
--uname   --uptime        # Yes, this calls ssh underneath!

so yeah i wrote a network transparent tab completion tool and option parsing library called textfree86

# ? Mar 17, 2018 08:42

Carthag Tuek: Oct 15, 2005; Tider skal komme,
tider skal henrulle,
sl�gt skal f�lge sl�gters gang

1) excellent name

2) in that last example, does it nest after the first command? its hard to tell what machine the commands are on because im dumb

# ? Mar 17, 2018 08:45

tef: May 30, 2004; -> some l-system crap ->

Krankenstyle posted:

1) excellent name

quote:

2) in that last example, does it nest after the first command? its hard to tell what machine the commands are on because im dumb

the last example is ./textfree86.py 'run this ssh command' -- 'then pass these arguments to it'

# ? Mar 17, 2018 13:50

Ator: Oct 1, 2005

tab completion is so helpful

# ? Mar 17, 2018 15:45

rjmccall: Sep 7, 2007; no worries friend; Fun Shoe

can�t you just do that completely in bash with complete -F and COMP_WORDS

# ? Mar 17, 2018 16:05

Thermopyle: Jul 1, 2003; ...the stupid are cocksure while the intelligent are full of doubt. �Bertrand Russell

rjmccall posted:

can�t you just do that completely in bash with complete -F and COMP_WORDS

I have no idea.

But this post made me think of all the times I developed some cool and complex system and then someone goes "can't you just do this completely obvious already-developed thing that if you had been better at research you'd have known about?"

and then i go "well poo poo"

# ? Mar 17, 2018 17:36

TimWinter: Mar 30, 2015; https://timsthebomb.com

Writing your own version of something existent gives it unique flavor, a new take on an old standard. Like a carefully crafted loss edit.

# ? Mar 17, 2018 18:02

Symbolic Butt: Mar 22, 2009; (_!_); Buglord

TimWinter posted:

Writing your own version of something existent gives it unique flavor, a new take on an old standard.

yeah that sounds cool and good--

TimWinter posted:

Like a carefully crafted loss edit.

oh no

# ? Mar 17, 2018 18:26

Adbot: ADBOT LOVES YOU

# ? Jun 1, 2024 04:21

tef: May 30, 2004; -> some l-system crap ->

rjmccall posted:

can�t you just do that completely in bash with complete -F and COMP_WORDS

i mean, it does use COMP_POINT and COMP_LINE because i wanted the setup to be 'add this line to complete' not 'download a bash function script and install it'

like i feel like this question comes from maybe not seeing what it's doing, but i guess my point isn't well explained

the point is that if i change the script on the remote end, the auto-complete suggestions change on the client

without editing my bash config settings or updating textfree86.py

# ? Mar 17, 2018 19:58

The Something Awful Forums > Discussion > Serious Hardware/Software Crap > YOSPOS > p-lang thread: (now (have you (problems two)))

«‹›1784 »