beneroth changed the topic of #picolisp to: PicoLisp language | The scalpel of software development | Channel Log: https://libera.irclog.whitequark.org/picolisp | Check www.picolisp.com for more information
ygrek has quit [Remote host closed the connection]
<abu[7]> Hi tankf33der!
<tankf33der> Hi
<abu[7]> Can't it be simpler? Do you flatten only the top level?
<abu[7]> Otherwise 'fish'
<tankf33der> Only top
<abu[7]> You could (or (pair L) (list L))
<tankf33der> I am building curl call list and supporting -u credentials only if passed
<tankf33der> I have built picolisp library for monero-wallet-rpc daemon. Very flexible. Now i could do whatever i want to.
<tankf33der> Simple example of usage
<abu[7]> cool
<tankf33der> i like this
<abu[7]> Yeah, but this goes deeper (not only top level)
<tankf33der> i have maximum 1 level: ("1" ("2" "3") "4")
rob_w has joined #picolisp
<abu[7]> ok
rob_w has quit [Ping timeout: 252 seconds]
<tankf33der> abu[7]: here or meeting ?
<abu[7]> Meeting in 2 h
<tankf33der> ok. no hurry
<abu[7]> No problem. A difficult question?
<tankf33der> can not make a array of objects from @lib/json.l
<abu[7]> 'printJson' ?
<tankf33der> yeap
<tankf33der> {"id": "0", "params": {"dests": {"amount": 1, "address": "aaa"}}}
<abu[7]> hmm, an older version had an "array" flag arg, but this is gone
<tankf33der> dests must be array
<tankf33der> array of objects..
<abu[7]> yeah
<abu[7]> works only for atoms
<abu[7]> str or num
<tankf33der> damn. eh.
<abu[7]> I don't remember why this was removed
<tankf33der> this is bad for me.
<abu[7]> We should revert
<abu[7]> it was somewhen in pil64
<abu[7]> still in latest pil64
<tankf33der> yes, please. i will use it
<abu[7]> Can you try with json.l from 24nov16abu
<abu[7]> if it makes sense?
<tankf33der> doing
<abu[7]> Do you have it?
<abu[7]> or should I paste?
<tankf33der> i will copy from github's repo
<abu[7]> ok
<abu[7]> Can you investigate if we can revert?
<abu[7]> import some fragments?
<abu[7]> The 'Arr' and 'T' stuff
<tankf33der> jsonOLD is beaking everything. new library - new knowledge how to use it. jsonrpc returns "Parse error" out of box
<abu[7]> oh
<abu[7]> That's why I say we need to import fragments
<abu[7]> the other parts are ok
<abu[7]> I don't remember why we removed it
<abu[7]> was not satisfactory somehow
<abu[7]> For printJson it just needs ((=T (car Item)) and (cdr Item)
<abu[7]> Shall I re-introduce into pil21?
<tankf33der> yeap. try
<abu[7]> Why was this feature removed?
<tankf33der> its not me.
<abu[7]> I hope it does not conflict
<tankf33der> do you use it somewhere ?
<tankf33der> you can test firstly
<abu[7]> I newer used the array feature iirc
<tankf33der> this code breaks anyway with old json
<tankf33der> i imagine "address" will be array of objects
<abu[7]> let me re-import Arr
<tankf33der> ok
<abu[7]> needs alsp parse
<abu[7]> packJson I mean
<abu[7]> This is a version without changes to packJson yet: http://pb1n.de/?a3d8a7
<abu[7]> Can you try?
<tankf33der> doing
<tankf33der> i will named jsonNEW
<tankf33der> i will named jsonNEW.l
<abu[7]> ok
<tankf33der> not breaking current code
<tankf33der> good sign
<tankf33der> and where i should put T ?
<abu[7]> Cons it to a list that should be printed as array
<abu[7]> OK, I also adjusted 'packJson'
<tankf33der> how to modify this code to make dests an array
<tankf33der> ?
<abu[7]> (list T '(id . "0") (list 'params ..
<tankf33der> hm. now working. checking
<tankf33der> [{"5927829442663": !? (cdar X)
<tankf33der> id -- List expected
<tankf33der> ?
<abu[7]> Hmm, it should work as before
<tankf33der> i never use json until this week.
<tankf33der> and never touched in pil too
<abu[7]> Can you investigate more?
<tankf33der> i will try, but you see on your side ?
<tankf33der> run this code with newer version
<tankf33der> with (list T ...
<tankf33der> works ?
<abu[7]> yes
<abu[7]> {"id": "0", "params": {"dests": {"amount": 1, "address": "aaa"}}}-> "}"
<tankf33der> with list T ?
rob_w has joined #picolisp
<abu[7]> It has no T
<abu[7]> I insert
<tankf33der> copy paste your code
<tankf33der> please
<abu[7]> I get an error
<abu[7]> Item is (id . "0")
<abu[7]> We should now have the same as old pil
<tankf33der> pil64 ?
<abu[7]> is (id . "0") wrong syntax?
<abu[7]> I don't remember
<tankf33der> i thought now.
<tankf33der> (cons 'id "0") or '(id . "0")
<tankf33der> i thought yes, correct
<tankf33der> but i do not have much correct examples. i reverse engineered library to understand how it is working. :(
<abu[7]> The code of ((=T (car Item)) and (T is different
<abu[7]> (printJson (car X)) vs (printJson (cdar X))
<abu[7]> and (prin "\"") etc.
<abu[7]> So old version was not right?
<abu[7]> Or different meanang?
<tankf33der> i can not use it with T
<tankf33der> jsonNEW.l i meant
<abu[7]> of course
<abu[7]> but what is correct?
<abu[7]> should the cond clauses be the same?
<abu[7]> ok, I change
<tankf33der> i am on jsonNEW.l right now.
<tankf33der> and need support for arrays.
<tankf33der> (list T 'dests ...)
<abu[7]> Same logic
<tankf33der> doing
<tankf33der> will be jsonNEW.l on my side
<abu[7]> for both clauses
<abu[7]> needs also packJson
<abu[7]> no
<abu[7]> I will release
<abu[7]> if ok
<tankf33der> code is working out of box so far
<tankf33der> and how to enable arrays on jsonNEW.l ?
<abu[7]> ?
<tankf33der> or what i testing ? :)
<abu[7]> Code for both clauses is redundant now
<abu[7]> should rewrite as 'if'
<abu[7]> moment
<abu[7]> no
<abu[7]> I keep it
<abu[7]> How about this? http://pb1n.de/?207b52
<abu[7]> wait
<abu[7]> Sorry, no time
<tankf33der> ok
<abu[7]> Can you investigate more?
<abu[7]> The correct way for print and pack?
<tankf33der> arrays ?
<abu[7]> all
<abu[7]> I want to release then
<tankf33der> eh. this is beyond me.
<abu[7]> No, you understand that
<abu[7]> I can't concentrate atm
<abu[7]> phone calls all the time
<tankf33der> doing
<abu[7]> and meeting soon all afternoon
<abu[7]> Thanks!! ☺
<tankf33der> no hurry
<abu[7]> yeah
<abu[7]> I think the code of json.l is no magic
<tankf33der> fucking recursion, this is my weak spot
<abu[7]> I wll look at it this evening. Perhaps the whole thing needs to be done differently
<abu[7]> parseJson and readJson are OK?
<tankf33der> seems so
<abu[7]> good
<tankf33der> i found json testsuite from Nicolas Seriot
<tankf33der> another now.
<abu[7]> Wait
<abu[7]> I made a stupid erro
<abu[7]> r
<tankf33der> ok
<abu[7]> replaced map with for
<abu[7]> moment
<tankf33der> ok
<tankf33der> much better.
<tankf33der> old code works. array works.
<abu[7]> cool
<abu[7]> Can you check more? Also the other functions? Then we can release this evening or tomorrow
<tankf33der> if that then i must write my own testsuite.
<tankf33der> lunch break.
<tankf33der> afk
<abu[7]> ok :)
<abu[7]> a test suite would be fantastic
<tankf33der> But it will test only parser
<tankf33der> readJson right?
<abu[7]> packJson is analog to printJson
<tankf33der> Why two then?
<abu[7]> ?
<tankf33der> They do different things, right?
<abu[7]> pack vs. print
<tankf33der> Aaaa, ok
<tankf33der> I am hacking from 4am, this is crazy
<abu[7]> oh! :)
<tankf33der> ret
<tankf33der> open this link
<tankf33der> you will see json below as example
<tankf33der> destinations is array of objects
<abu[7]> Later, now in meeting
<tankf33der> not 1,2,3 BUT pairs
<tankf33der> of course.
<tankf33der> testsuite will be here
<tankf33der> but first you should not use (quit ...) :)
<aw-> tankf33der: i should run my parser against those files
<tankf33der> :)
<aw-> all passed
<tankf33der> perfect
<aw-> my parser returns NIl if it fails to parse a JSON string (or file)
<aw-> i wrote a quick script to test each file and i just got a large list of NILs haha
<aw-> i dunno if that's correct
<aw-> it doesn't treat a lonely "true" as a valid JSON
<aw-> even though technically it is
<aw-> i find it stupid
<beneroth> booleans are easy, the hard is the allowed E-notation for numbers (even such which are not valid in Javascript)
<beneroth> I guess it is not widely used in practice, but still...
rob_w has quit [Remote host closed the connection]
<tankf33der> current json can not create such json: {"mike": [{"1": "2", "2": "2"}, {"1":"2","3":"4"}]}
<tankf33der> this can: {"mike": [1, 2]}
ygrek has joined #picolisp
<abu[7]> ret
<abu[7]> tankf33der, I think 'quit' is needed. Can you 'catch' it?
<tankf33der> But i fail on faulty tests
<abu[7]> You cat test '@@' after catch
<abu[7]> See the 'if2' example in the ref for 'catch'
<tankf33der> i do (in "fail.file" (readJson))
<tankf33der> right ?
<abu[7]> K
<abu[7]> T
<tankf33der> having fun
<abu[7]> 👍☺
<abu[7]> Instead of (in "file" ..) you can also use 'pipe' or 'input'
<tankf33der> catch ok
<tankf33der> lets see
<abu[7]> : (let S (chop "{a: 1}") (input (++ S) (readJson)))
<abu[7]> -> ((a . 1))
<tankf33der> if fail file passed is it ok ?
<abu[7]> I think this is good. A negative-test.
<tankf33der> pass files passed
<tankf33der> for one test
<tankf33der> doing second
<abu[7]> What is your impression? Shall I release it?
<tankf33der> you can release it, but i did not get what i really need. see above.
<tankf33der> failed, but should somehow pass
<abu[7]> Which above?
<tankf33der> 17:02 <tankf33der> current json can not create such json: {"mike": [{"1": "2", "2": "2"}, {"1":"2","3":"4"}]}
<tankf33der> 17:03 <tankf33der> this can: {"mike": [1, 2]}
<abu[7]> How did you make the second?
<abu[7]> (printJson '((mike (T (a . 1)))))
<abu[7]> (printJson '((mike (T (a . 1)))))
<abu[7]> oops
<tankf33der> (printJson
<tankf33der> (list
<tankf33der> (list 'mike 1 2) ) )
<tankf33der> like this
<tankf33der> and this code does not work on jsonNEW.l
<tankf33der> but jsonNEW passed test suite
<tankf33der> no regressions
<abu[7]> Hmm, should we redesign the whole lib?
<tankf33der> eh. hard decision
<abu[7]> Is it only print?
<abu[7]> or also read?
<abu[7]> (printJson '((mike (a . 1)))) {"mike": {"a": 1}}
<abu[7]> correct?
<abu[7]> (printJson '((mike ((a . 1) (b . 2))))) {"mike": {"(a . 1)": {"b": 2}}}
<abu[7]> This too?
<tankf33der> second is not correct
<tankf33der> imho
<abu[7]> true
<tankf33der> thats why i always use list or cons
<tankf33der> bulletproof
<abu[7]> Thats the same
<tankf33der> of course, i am a human :)
<abu[7]> quoted
<abu[7]> So let's not release it
<abu[7]> I must rethink the whole structure
<abu[7]> I don't even know a BNF for Json
<tankf33der> there is a picture
<abu[7]> ok, I think about it tomorrow
<abu[7]> Too bad that this page does not work in a text browser
<tankf33der> afk
<abu[7]> o/
<abu[7]> me too
<tankf33der> json BNF in txt
<abu[7]> The problem is unsolvable
<abu[7]> The s-expr's we have don't map uniquely to all possible combinations of Json syntax
<abu[7]> We would need to define more markup syntax as we have with 'T'
<abu[7]> I find tomorrow ;)
<abu[7]> A possible subset: http://pb1n.de/?696dc8
<abu[7]> Gives these possibilities: http://pb1n.de/?05bed5
<abu[7]> Good night
<beneroth> yeah, I made a DOM (similar to HTML DOM) from JSON based on pil symbols/OOP, and then have methods to print/pack it, functions to parse JSON and generate the DOM, and a DSL to generate the DOM
<beneroth> somewhat overhead, but can represent and handle all JSON
<aw-> hello
<aw-> I wrote a JSON parser almost a decade ago for PicoLisp
<aw-> why are people in here ALWAYS trying to re-create the same thing?
<beneroth> lisp curse
<aw-> this has been a solved problem, it works
<beneroth> (I need support for true)
<bjorkintosh> aw-: NIH, or lisp curse as beneroth said..
<beneroth> NIH and lisp curse is not the same thing I would say, though arguably highly related
<bjorkintosh> you're right.
<bjorkintosh> it's NIH NIMBY Lisp curse.
<beneroth> well I wrote my own JSON lib because I wanted full support also for untrusted and/or partially broken JSONs.
<beneroth> I've also addded a way to specify the expected JSON structure so the parser can stop early when the JSON structure is not in the expected format
<beneroth> my use case is interfaces/imports/APIs of bullshit business software
<aw-> lisp curse, holy hell
<aw-> beneroth: JSON-schema?
<beneroth> no, I've looked into it but couldn't make me commit to that mess yet xD
<beneroth> aw-, I think the reason I didn't use your (very good) implementation is because I needed support for floating point / decimal numbers
<aw-> yeah that's a hairy mess I'd rather avoid in PicoLisp haha