<freemint>
Not as far as i know. But the hutter price a time sink with no bottom
<razzy>
it is, it is also good excecise
<freemint>
the literal go of the hutter price is to make a general artificial intelligence
<freemint>
It is beyond the scope of a masters thesis in the field of data compressions
<razzy>
yes :]
<freemint>
people have clusters of computers running, trying different compression variations
<razzy>
think so
<freemint>
1. you are misusing picolisp for what it is not intended
<razzy>
it also means it is dataset with lots of experimental data
<razzy>
and comparison
<freemint>
2. It is not an good exercise to learn PicoLisp since PicoLisp is for other thingd
<razzy>
like?
<freemint>
Developing web apps?
<razzy>
it is also low lvl lisp
<freemint>
Building abstractions to make stuff easier
<freemint>
It is not as low level as you think.
<freemint>
It is simple, but if it were low level you could modify the free-list
<razzy>
and it is awesome for low lvl comprimation and AI research
<freemint>
if it were low-level you could implement arrays directly from Lisp
<razzy>
you can not?
<freemint>
not without compiling something to assemly
<freemint>
there is no gpu support
<freemint>
the days of eurisko are over!
<freemint>
Nearly no AI research happens in Lisp
<freemint>
SICP is no longer taught at MIT
<razzy>
yop
<razzy>
you are right in all cases
<razzy>
as far i know
<freemint>
Picolisp is awesome and i sometimes have thoughts similar to you. But that is nostalgia for a time where neither of us was (probably) alive
<razzy>
gn, i take you are not willing to help and i need to sleep
<freemint>
not if the goal is hutter
<freemint>
and there are better things to misuse picolisp for
<razzy>
like?
<razzy>
goal is to have adaptible compression
<freemint>
How about searching the computational universe for new groups
<razzy>
for everything
<razzy>
new groups?
<freemint>
do you know kolmogorov?
<freemint>
(mathematical concept from which you can build other stuff)
<razzy>
some of his work
<freemint>
Also if you need new ideas for approaches for hutter i can give you some. i got a list somewhere. but i will not spend any minute implementing something to do hutter
<freemint>
You are aware that n-bits of information on average can not be compressed to be less that n bits?
<razzy>
freemint: are you agains AI?
<razzy>
freemint: i know there is limit on compression
<freemint>
you are aware that finding out the true content of entropy of a structured thing is equivalent to the Entscheidungsproblem?
<freemint>
No, but i am very doubtfull of Hutters approach
<razzy>
finding information compression limit tells us something about our universe
<freemint>
really what does it tell us?
<freemint>
*Halteproblem
<razzy>
the essence of information, nature of stuff
<freemint>
I do not see why
<razzy>
nobody seen maximum comprimation yet
<freemint>
It tells us that sometimes more information can be broken down to be represented by less
<freemint>
Oh we have seen ideal compression
<freemint>
But not for general language.
<razzy>
only in stuff we created
<razzy>
when you find ideal compression, you found a way god/universe created it
<freemint>
why should there by only one ideal compression?
<razzy>
because you have least bits used
<freemint>
why can't there be multiple ideal compression producing the same length?
<freemint>
we are unable to store/process/see all information the universe contains
<razzy>
freemint: could be, very unlikely tho
<freemint>
why do you think so?
<freemint>
this is a very bold claim
<freemint>
a claim which can not be proven in ZFC set theory
<razzy>
because it is very very hard to find two transformations that has same starting lenght and same result
<freemint>
for a given set of data? That is easy
<razzy>
could be infinite possibilities
<razzy>
freemint: ok, that you could not compress more
<freemint>
but you can not say there can't be two
<razzy>
i just saying it is very unlikely
<freemint>
it is as unlikely as the existence of one optimal compression
<razzy>
and very unlikely is still infinite possibilities
<freemint>
are we even sure there is one?
<freemint>
The lemme of Zorn which might give you a minimum only work on compact sets
<razzy>
nope, we are trying to find some
<freemint>
*Lemma
<freemint>
ok, then i got another argument against doing it now
<freemint>
For any computation we do now we increase total entropy making calculations in the future harder even when computers are better
<razzy>
yes :D
<freemint>
when you goal is for humanity to find aixi is it unethical to try to find it now
<razzy>
but any burning tree makes so much more mess
<freemint>
so what? stop them from burning trees
<razzy>
:D nice,.. you got me there :D
<razzy>
burning tree makes so much more mess than computing computer
<freemint>
and even if that makes more entropy than you running a computer, you are still irresponsible if you burn more entropy than you have too
<razzy>
uff,.. i have 3am,.. i am going to sleep
<razzy>
freemint: agreed
<freemint>
sorry but aixi is a delusion in my opinion.
<razzy>
in what sense?
<freemint>
it is true when you run things to infinity
<freemint>
but when you run things to infinity there are as many positive even as natural numbers
<razzy>
last statement is false imho
<freemint>
This statement is true in the infinite case but false in the finite
<razzy>
nope
<freemint>
what part is false?
<razzy>
there is bigger and smaller ifinity
<razzy>
infinity
<freemint>
you are wrong i can construct a bijective mapping
<razzy>
show me
<freemint>
these infinities are definitely of the same size
<freemint>
i would map both bijectivly to the positive natural numbers
<razzy>
what are positive and what natural numbers?
<razzy>
in your view?
<freemint>
Positive and natural number are number which i can creat with zero and the successor function
<freemint>
*one
<freemint>
let me restate: There are as many positive even natural numbers as there are natural numbers
<freemint>
and the fact that |N has the same size as {successor(x)|x \in |N}
<freemint>
razzy ... i am right
<freemint>
oh an the inverse mapping is x to x/2 -1 and i can divide by 2 because all numbers are even
<razzy>
well,.. i need sleep
<razzy>
write back tomorow?
<razzy>
and i like mathematics graph
<freemint>
i can
<razzy>
i will write back tomorrow
singa has quit [Remote host closed the connection]
<freemint>
remember i only talked about natural, if i would have said real i would be false.
singa has joined #picolisp
<razzy>
i will check tomorrow
pierpa has quit [Quit: Page closed]
mist3 has joined #picolisp
mist3 has quit [Remote host closed the connection]
_whitelogger has joined #picolisp
pierpal has quit [Quit: Poof]
pierpal has joined #picolisp
Looking has joined #picolisp
Looking has quit [Read error: Connection reset by peer]
orivej has quit [Ping timeout: 244 seconds]
aw- has joined #picolisp
beneroth has joined #picolisp
<beneroth>
Good morning all
<Regenaxer>
Hi beneroth! Welcome back!
rob_w has joined #picolisp
<beneroth>
Thanks Regenaxer :)
mtsd has joined #picolisp
pierpal has quit [Quit: Poof]
pierpal has joined #picolisp
werkin has joined #picolisp
alexshendi has joined #picolisp
ArsenArsen4 has joined #picolisp
ArsenArsen4 has quit [K-Lined]
ddstreet20 has joined #picolisp
ddstreet20 has quit [K-Lined]
TReK6 has joined #picolisp
TReK6 has quit [Remote host closed the connection]
werkin has quit [Quit: Leaving]
werkin has joined #picolisp
<freemint>
Good Morning
<freemint>
wheew that debate was long ...
freemint has quit [Ping timeout: 240 seconds]
CGML23 has joined #picolisp
CGML23 has quit [Remote host closed the connection]
<Regenaxer>
yes
thurin29 has joined #picolisp
thurin29 has quit [Remote host closed the connection]
alexshendi has quit [Ping timeout: 268 seconds]
xificurC has joined #picolisp
conno21 has joined #picolisp
conno21 has quit [Read error: Connection reset by peer]
freemint has joined #picolisp
<freemint>
razzy https://youtube.com/watch?v=JjDsP5n2kSM watch the first 38 minutes of that talk but replace "less memory use" with "better compression ratio"
alexshendi has joined #picolisp
freemint has quit [Remote host closed the connection]
freemint has joined #picolisp
<freemint>
Regenaxer if i want to return stuff in a custom mime-type. Can i do that with http.l or do i need to rewrite 'http ? How about if do not know the mime-type until my own code runs, do i need to rewrite 'http then?
<Regenaxer>
No need
<Regenaxer>
See the wiki, it detects the mime type and stores it with te document
<freemint>
Also can i add additional header attributes depending on the request?
<freemint>
content with mime-type is generated (like rss stream)
<freemint>
(so it is not on disk,
<Regenaxer>
Still, it must be determined automatically
<Regenaxer>
The other question, see 'http1'
<beneroth>
even if @lib/http.l does not offer it, its easy to add, no need to rewrite whole 'http in any case
<Regenaxer>
For the mimetype, from wiki:
<Regenaxer>
(class +Bin +Entity)
<Regenaxer>
(rel nm (+String)) # Name
<Regenaxer>
(rel mt (+String)) # Mime type
<Regenaxer>
(dm upd> (X Old)
<Regenaxer>
(rel d (+Blob)) # Binary data
<Regenaxer>
(when (== X 'd)
<Regenaxer>
(put> This 'mt (mimetype (blob This 'd))) ) )
<Regenaxer>
beneroth, T
<Regenaxer>
But http1 is used for that, in several places
<Regenaxer>
eg 'httpHead'
<Regenaxer>
or just write it all directly
<Regenaxer>
nothing more than a hand full of 'prinl's
<freemint>
i do not see how that works with dynamic content
Georgyo8 has joined #picolisp
<Regenaxer>
I pil all content is generated dynamically
<freemint>
Truism
<Regenaxer>
in the sense of no static pages I mean :)
Georgyo8 has quit [Remote host closed the connection]
trisk17 has joined #picolisp
trisk17 has quit [Remote host closed the connection]
pierpal has quit [Quit: Poof]
<freemint>
i read the http code and i came to the conclusion that every time a !function is called an http header is generated.
pierpal has joined #picolisp
<Regenaxer>
yes, that's the way
<freemint>
ok on what URL can i have an rss stream then?
<freemint>
when it can not end in !rss?para=meter ?
<Regenaxer>
?
<freemint>
My plan was: i got an URL A which requires http authentication and if the authentication is valid (checked with dB) the user is displayed new private messages
<freemint>
the new private messages are pulled from his associated mailbox object from the database
<Regenaxer>
You can use Server-Sent Events to push messages to the user
<Regenaxer>
beneroth: Vacation over?
<freemint>
This can not use Server-Sent Events since it should be readable by an rss (or activitypub to be honest) client
<razzy>
freemint: the infinities are the same with natural 1,2,3,4, and even -4,-2,2,4,
<freemint>
yes i told ya so
<Regenaxer>
I don't understand exactly, but activitypub is a good thing
<freemint>
I can not talk with another activitypub server over HTML, i need to use json-ld.
<razzy>
freemint: you told "let me restate: There are as many positive even natural numbers as
<razzy>
there are natural numbers [02:48]"
<freemint>
these json-ld files need to be dynamicly generated
<Regenaxer>
perfect, then do so ;)
<Regenaxer>
JSon is almost trivial to parse and to print
<Regenaxer>
several ways (3)
<freemint>
how can i deliver dynamic json with http.l
<Regenaxer>
Just print
Dread19 has joined #picolisp
<freemint>
but then it got an http header with a wrong mime-type
<Regenaxer>
Sigh. Why?
<Regenaxer>
You get what you print
<Regenaxer>
it is up to you!
<Regenaxer>
Did you look at what is there, and how you can combine it?
Dread19 has quit [Remote host closed the connection]
<Regenaxer>
It boils down to
<Regenaxer>
(httpHead NIL Upd)
<Regenaxer>
(ht:Out *Chunked
<Regenaxer>
(prinl ...
<Regenaxer>
or use higher-level pre-cooked functions
<Regenaxer>
ie http1 instead of httpHead
<Regenaxer>
if httpHead is not enough
<freemint>
where would i call this code?
<Regenaxer>
In your program?
<freemint>
in a '!function?
<freemint>
in a .l file which is loaded? and executed?
<Regenaxer>
both are good
<Regenaxer>
equivalent
<Regenaxer>
Things after the '?' arrive as arguments then
<freemint>
"i read the http code and i came to the conclusion that every time a !function is called an http header is generated." to which you said that's the way. implies that i can not modify the header once i am in a function called over the ! mechanism
<Regenaxer>
no, the function *generates* the header
<Regenaxer>
so you check first, then decide which header to print
<freemint>
oh so the function is given the raw socket for a moment?
<Regenaxer>
the current output stream goes to the client of this HTTP transaction
<Regenaxer>
and the input stream too in case it is needed
<freemint>
ok ... i am sorry that i wasted your time on this question
<Regenaxer>
look eg at wiki/lib.l
<Regenaxer>
(I don't know if you have access to other examples)
<Regenaxer>
getBlob for example
<freemint>
i can not find get blob but should download, and pdf do as examples?
<Regenaxer>
The first function in wiki/lib.l ?
<Regenaxer>
Not sure if it is a good example
<freemint>
mhh outdated source code again
<Regenaxer>
oh, but this is very old I thought
<Regenaxer>
ok, well, 19jun17abu
<freemint>
02may15abu
<Regenaxer>
seems it was called "download" before
<freemint>
Picolisp Really needs a central repo
<Regenaxer>
So you have 'download'
<Regenaxer>
It is, software-lab.de
<Regenaxer>
always up to date
alexshendi has quit [Ping timeout: 240 seconds]
<aw->
freemint: i have a code sample for your exact use case
<freemint>
It is a json based protocol which servers can implement to communicate with other servers or clients in a federated way to implement message passing functions for example for social networks distributed over many servers
<aw->
i've seen this before.. a JSON document with a bunch of urls to other JSON documents..
<freemint>
(i was just changing trains)
<aw->
ah yes, JSON-LD
<aw->
this looks just like json-ld
<freemint>
it is jsonld
rob_w has quit [Quit: Leaving]
<freemint>
but a protocol based in json-ld which implements some basic functions and tells the servers how to behave
<freemint>
it is interesting since many open source social networks implemented it. Examples are peertube, mastodon, Gnu social
<aw->
right
<aw->
i have no interest in social networks, so it's not interesting to me
<aw->
i looked into this a few months ago actually but couldn't find a valid use case
<freemint>
this is an interesting claim
<aw->
?
<beneroth>
so.. its re-inventing usenet? xD
<aw->
why is the JSON-LD "documentation" website only linking to YouTube videos?
<beneroth>
aw-, because modern people can't read xD
<aw->
"What is JSON-LD? - watch this video!"
<aw->
fuck
<aw->
beneroth: yeah
<mtsd>
Hello everyone
<mtsd>
aw-, I got the book you recommended, "Elements of computing systems"
<aw->
mtsd: hi! awesome!!
<aw->
did you start it?
<mtsd>
Seems very good, I only read the first chapter so far
<mtsd>
Busy summer
<mtsd>
Thanks again for the recommendation!
<freemint>
the json-ld documentation sucks.
<beneroth>
btw books, I'm reading "Designing Data-Intensive Applications", gives a good overview about different database concepts. Funny to read with an understanding of pilDB.
<aw->
mtsd: no problem, it's lots of fun, i love that book. Don't forget to do the exercises when you have a chance, it's really worth it
<mtsd>
I will!
<freemint>
i agree that it drives on some academic hype ways. like the semantic web and so on, but there is an interesting community using activity-pub based software
<aw->
freemint: JSON-LD abuses JSON. It's not good.
<beneroth>
freemint, semantic web is a "coming-soon"-hype for like 10 years now. it's BS.
<aw->
the fact that you need to parse the document in a special way is a big fail, in my opinion
<freemint>
beneroth i am aware that it is bullshit. that's why i wrote that disclaimer
<aw->
JSON is a great format because it's super simple, easy to implement and parse. JSON-LD adds a layer of complex on top of an already parsed document, for what? so you can get links to more JSON-LD documents for even more parsing...
<freemint>
aw- fun thing most activitypub implementations ignore the JSON-LD part
<beneroth>
I don't think semantic web dream can be implemented, not without a central database of term definitions. without it, all "publishers" of information use different terms differently, or the same terms using different stuff. so much context-dependence. and the frame of reference is constantly changing and evolving (red-queen hypothesis applies, I think).
<aw->
freemint: if "most" ignore it, doesn't that break the spec? or leave room for ambiguity?
<beneroth>
aw-, I think if one is designing a new protocol on top of JSON is kinda throwing away the good parts of JSON while keeping the bad parts (e.g. number representation)
<beneroth>
it seems like a "golden hammer tool" to me. people love it so much they want to use it for everything without sense.
<freemint>
beneroth if you call "red queen" what adversarial competing agents do you see in information Representation?
<beneroth>
the constantly evolution of language and human knowledge?
<freemint>
aw- well as long as you only act on @contexts you have implemented, you are fine and if recall you can demand that all stuff send to you in contexts you implemented
<aw->
hmmm
<aw->
MQTT is designed exactly for this use-case
<freemint>
so basically it requires things in the spec which are not really needed - which makes it a bad spec
<freemint>
but in picolisp we should be flexible enough to handle @context properly
<freemint>
in the long run i am after the community which is not these platforms
<aw->
yes at least it's actually valid JSON, so i can't complain about that
<aw->
but since ActivityPub is specifically for social networking, it's not something i looked into
<freemint>
the whole activity-pub thing is designed so that all notifications from all the websites you use (and support activity-pub). can be funneled into one strea$
<aw->
if you're just receiving a JSON string, just use json.l to parse it and (cdr (assoc "@context" Jsondata))
<aw->
doesn't seem difficult to implement support for ActivityPub in PicoLisp
<freemint>
one of the usecases is that blender can host there own peertube instance when youtube blocks them by accident again.
<freemint>
and you can get "new video uplaod" notification without having to check there website. you just have to check your mastodon instance
<freemint>
*their
<freemint>
so in that sense it is kinda like an server collecting rss streams with the additional feature that you can comment and "like" things and people in your social circle can see it.
<aw->
if you host your own ActivityPub API, does it get notifications?
<freemint>
well the API has 3 parts
<freemint>
client -> server, server -> client and server <=> server
<freemint>
if you implement server <=> server you can get notifications from other servers
<freemint>
(and tell them you liked there content ... and so on)
<freemint>
aw- does that address your question?
<aw->
ok thanks
pierpal has quit [Quit: Poof]
pierpal has joined #picolisp
Zuu_ has joined #picolisp
Zuu_ has quit [Remote host closed the connection]
mtsd has quit [Quit: Leaving]
freeemint has joined #picolisp
freeemint has quit [Client Quit]
freeemint has joined #picolisp
freeemint has quit [Client Quit]
freemint has quit [Ping timeout: 260 seconds]
freemint has joined #picolisp
freemint has quit [Ping timeout: 240 seconds]
werkin has quit [Quit: Leaving]
freemint has joined #picolisp
<beneroth>
bbl
beneroth has quit [Quit: Farewell]
Cronus7 has joined #picolisp
Cronus7 has quit [Remote host closed the connection]
Cool_Fire26 has joined #picolisp
Cool_Fire26 has quit [Remote host closed the connection]
<freemint>
aw- what is on-prem for exactly?
<freemint>
(or roughly)
<freemint>
Regenaxer, have you thought about implementing 'app another way. The "each session a port" idea introduces problems like links dying, and makes it hard for search engines
<Regenaxer>
Yes, in the beginning, but I think it is the optimal way
<Regenaxer>
What links are dying4
<Regenaxer>
?
<Regenaxer>
What is needed is one process per session. So an individual port is perfect
<Regenaxer>
How else would you do it?
<freemint>
well we got lab.de/5324/static/pic.png
<freemint>
(for example)
Odd_ has joined #picolisp
<freemint>
and when some time passes this url "dies" and becomes unavailable with no graceful way to recover/redirect to login page
<Regenaxer>
No, it never timeouts as long as a client is connected
<Regenaxer>
And without client the session is meaningless
<Regenaxer>
It is all about the context of the session
Odd_ has quit [Killed (Sigyn (Spam is off topic on freenode.))]
<freemint>
ok and i think the context of a session is not always appropiate.
<Regenaxer>
You can operate without sessions too, then no port is allocated
<Regenaxer>
Again, see the wiki. It is without session until you log in
<freemint>
Let's assume the wiki has private pages which can be only seen when logged in. How would i refer you this private page when you are logged in you can acces it and when not you presented with a login screen to use your credentials
<Regenaxer>
I have such things in an application in fact
<Regenaxer>
It uses long keys of the member in the URL
<Regenaxer>
It is not so secret for this test page: https://7fach.de/cgf/!landing.html?+13121734705783290
pierpal has quit [Quit: Poof]
<Regenaxer>
Static page, visible only for one who has this key
pierpal has joined #picolisp
<freemint>
and how can i generate such a static page?
<Regenaxer>
Easy, just the normal way
<Regenaxer>
These pages are generated with the standard pil gui
<Regenaxer>
on the fly
<Regenaxer>
no static page involved
<freemint>
the normal way is to copy the URL
<Regenaxer>
hä?
<Regenaxer>
what kind of copy?
<freemint>
if want to give somebody the link to my private Facebook profile i go to my profile and STRG + C the url
<Regenaxer>
yes
<freemint>
this URL is in a nice format too
<Regenaxer>
I just did that with the above url
<freemint>
and there are no different URLs pointing to the same content
<Regenaxer>
What do you want?
<Regenaxer>
except for complaining
<freemint>
Honestly i wanted to understand what is possible under http.l
<Regenaxer>
good, yes
<Regenaxer>
I see no limits
<freemint>
i am sorry if i have an antagonizing tone very often
<Regenaxer>
ok, no worry
<freemint>
The thing is "non"-human guessable URLs hurt discoerability
<freemint>
anyway how do other places do session management?
<freemint>
I think they put the session state in a datastructure and load the appropriate session via cookie information
<freemint>
i see that the PicoLisp approach is better as it keeps things secure
<Regenaxer>
"non"-human URLs like the landing.html are necessary because some auth is needed
<Regenaxer>
They are never typed manually, but communicated to the members
<freemint>
opioin please : What do you think about using the database for session storage and have a process each connection which gets the relevant state from the database?
<freemint>
(indexed by cookie
<Regenaxer>
Yes, this might be good for some purposes
<Regenaxer>
But I would not want to depend on cookies
<Regenaxer>
What if you need to open 2 sessions to the same application?
<Regenaxer>
I have that sometimes
<freemint>
would picolisp (the DB) take care that all connections which share a cookie have the same dB state?
trqx5 has joined #picolisp
<Regenaxer>
There are no mechanisms for that currently
<freemint>
well when you want different sessions use the private mode would be the workaround which has it's own problems
<freemint>
maybe there is a way to configure cookies such that they stay tab local - no idea!
<Regenaxer>
Same here
<freemint>
i will add that to my queue
trqx5 has quit [Read error: Connection reset by peer]
<freemint>
when i find something i tell you
<Regenaxer>
Yes, try
<freemint>
this website you sent me was really polished
<Regenaxer>
Design is not by me of course ;)
<freemint>
was it for some conference
<Regenaxer>
No, it is permanent. Ongoing projects
drathir5 has joined #picolisp
<Regenaxer>
hmm, seems I posted too much.
<freemint>
ok what for?
<Regenaxer>
It has more than just the empty test project
drathir5 has quit [Remote host closed the connection]
Metacity9 has quit [Remote host closed the connection]
<razzy>
only real way to better life (without exploiting anyone) is to do things more efficiently
<razzy>
like consumer goods forum
<razzy>
btw, is there library in picolisp for web interface? download page, ask for data on standard API interfaces
<freemint>
yes there a standard web interace.
<freemint>
It is included in html.l and http.l
<freemint>
There are several examples there is the wiki, the "app" (in the PicoLisp root folder) and the family example (that one is zipped in @lib/ i think
<freemint>
razzy the documentation however is sparse as always
<freemint>
razzy family is in @doc
pierpal has quit [Quit: Poof]
pierpal has joined #picolisp
Xenogenesis20 has joined #picolisp
Xenogenesis20 has quit [Remote host closed the connection]
montag45114 has joined #picolisp
montag45114 has quit [Remote host closed the connection]
darkmagic has joined #picolisp
darkmagic has quit [Remote host closed the connection]
<razzy>
i like small efective doccumentation
daemon4 has joined #picolisp
daemon4 has quit [Remote host closed the connection]
Alistair28 has joined #picolisp
Alistair28 has quit [Killed (Sigyn (Spam is off topic on freenode.))]