cess11_ has quit [Remote host closed the connection]
aw- has quit [Ping timeout: 240 seconds]
aw- has joined #picolisp
<Regenaxer>
aw-, what is wrong with the line messenger? My wife uses it a lot with her relatives in Japan
<aw->
hi Regenaxer
<aw->
they recently changed their privacy policy
<aw->
yesterday
<Regenaxer>
oh
<aw->
if you accept it, you are giving them permission to share (with other companies) your Timeline, profile photo, stickers/stamps usage, and lots of statistics about how you communicate with people
<aw->
the only thing that remains "private" (apparently) are your messages and photos
<aw->
but **everything else** defaults to shared with companies - for advertising/$$$, and you can't opt-out
<Regenaxer>
hmm, I see
<aw->
it's really bad
<aw->
it's bad because everyone in Japan uses LINE
<aw->
if you don't have LINE, it's like you don't exist
<Regenaxer>
yes, I know
<aw->
that's why i asked "how do we move everyone away from it?" - it seems like an impossible task
<Regenaxer>
seems so
<aw->
i suggested Wire because their stack, as well as client apps, are 100% open source
<tankfeeder>
wire are authors of skype
<Regenaxer>
Open source is not the problem, but who controls the metadata
<tankfeeder>
what about telegram ? :)
<aw->
yes exactly
<Regenaxer>
I mean, opes source does not fix it
<Regenaxer>
bitmessage
<Regenaxer>
We should write pilMessage
<aw->
but Open source at least provides you some confidence in the protocol/software
<aw->
much more than 100% closed system
<Regenaxer>
true, but what is interesting is the metadata
<aw->
tankfeeder: i used telegram for a while, but their server/backend is not open source
<Regenaxer>
Metadata give you control and info about people
<aw->
yes to an extent, I agree, but it's much less info than your full conversation history/photos/etc
<Regenaxer>
T
<aw->
metadata can be useful to link people, make assumptions about their location, etc
<aw->
oh Regenaxer funny story for you
<Regenaxer>
and, most of all, political, religious etc orientation
<aw->
i was at a cafe a few weeks ago with a friend, and we were arguing about (assoc)
<Regenaxer>
don't need contents for that
<aw->
then i opened PilBox on my Android and we resolved it hahaha
rob_w has joined #picolisp
<Regenaxer>
haha, cool! :)
<tankfeeder>
aw-: if your "problem" open source of every part of stack, than my "problem" signal protocol.
<aw->
i don't know much about Signal protocol
<Regenaxer>
I have an idea how we could write pilMessage
<aw->
does Wire use Signal protocol?
<tankfeeder>
aw-: i think so
<tankfeeder>
everybody use it except telegram
<Regenaxer>
One problem (besides scalability) of bitMessage is that it does not run on mobile devices
<Regenaxer>
pilMessage would run on PilBox :)
<Regenaxer>
Have some ideas about it
<tankfeeder>
or riot messenger+ matrix server or telegram secure chats
<aw->
tankfeeder: signal protocol is not good?
<tankfeeder>
aw-: its ok and for nsa too
<aw->
tankfeeder: link?
<aw->
do you have a source for that? or just writing based on your opinion?
<tankfeeder>
i will not find links fast
<tankfeeder>
because of protocol superusers can read our chats in whatsapp and any signal based IM
<tankfeeder>
can -> could
<aw->
tankfeeder: even if you run your own servers?
<Regenaxer>
and, most of all, needs a central server
<Regenaxer>
this is the core problem!
<Regenaxer>
metadata!
<Regenaxer>
We need PilMessage
<Regenaxer>
peer to peer
<beneroth>
there are TOR-based chat-systems now which are supposed to be good. dunno if scalable and practical.
<Regenaxer>
ok
<beneroth>
Regenaxer, agree
<anjaa>
don't you mean pil-to-pil?
<anjaa>
/joke
<Regenaxer>
tor comes close probably
<anjaa>
1
<Regenaxer>
anjaa, yep :)
<cess11>
how about tox?
<beneroth>
never heard of Line. so this is a general trend, that whole populations use a single chat system which becomes the infrastructure for much (incl. economic and gov) communication ?
<beneroth>
China made the WeChat-ID usable as passport-like identifier recently
<beneroth>
clacke[m] uses matrix I think, even in here.
<tankfeeder>
beneroth> never heard of Line.
<tankfeeder>
i cant belive it
<beneroth>
heard of WeChat?
<tankfeeder>
the most monetized IM ever, very success in business
<tankfeeder>
beneroth: wechat the same
<beneroth>
this stuff is easy & comfortable. people love that. and gov loves it for gaining an attack vector into every social/economic interaction
<aw->
beneroth: the problem with LINE is that everyone uses it, the great thing about LINE is that everyone uses it.
<beneroth>
aw-, aye, I see. network effect.
<beneroth>
too bad we (as humans) cannot be more rational all have a XMPP/jabber server at home
<aw->
yep
<aw->
i run an XMPP server
<aw->
works perfectly
<beneroth>
you use Jitsi? its a good skype-replacement
<aw->
no
<aw->
ejabberd
<beneroth>
ok
<aw->
+ Conversations on Android
<tankfeeder>
testing pil32 on aix with grid crash
<Regenaxer>
good
<aw->
beneroth: i've updated my JSON lib to successfully parse scientific numbers, as well as floating-points. Didn't push the code yet
<aw->
{"test":12.345e+32} ... like that -> (("test" . 12.345e+32))
<beneroth>
aw-, ok. thanks for your effort and initiative :)
<beneroth>
though I don't have a need/use case for this yet :)
<aw->
i'm not sure if Regenaxer will approve my method
<Regenaxer>
Why not?
<beneroth>
aw-, I would apply "it is good when you cannot take more away from it" (reducing it more)
<aw->
I use (intern)
<aw->
(intern (pack "1." 345 "e+" 32))
<Regenaxer>
Why is that necessary?
<aw->
maybe it's an ugly hack to prevent (str) from turning it into an integer
<Regenaxer>
clobbers the symbol space
<aw->
yes exactly
<Regenaxer>
I almost never use 'str', there are usually direct ways
<Regenaxer>
Can't you simply use 'native' to parse the number?
<Regenaxer>
would also be fastest
<Regenaxer>
ah, pil32?
<aw->
haha yeah
<aw->
well.. i will try this
<aw->
i'm not aiming for speed, but accuracy
<tankfeeder>
aix works, gcc 4.8.2
<tankfeeder>
this is good.
<aw->
i want it to perfectly parse JSON strings according to spec, even if it's "slower"
<Regenaxer>
If a compiler is available at 'load' time, you could use inline C then
<Regenaxer>
otherwise a shared lib
<Regenaxer>
Yeah, understand
<tankfeeder>
now solaris 10, sparc, gcc 3.4.3 :)
<Regenaxer>
aw-, another option would be (in '(externalparser ...))
<beneroth>
aw-, on which picolisp version are you?
<Regenaxer>
iirc there is a float equivalent to 'bc' (?)
<beneroth>
tankfeeder, good that you keep hunting this
<Regenaxer>
yes, indeed, this is a disturbing problem
<tankfeeder>
gcc 3.4.3 crash
<tankfeeder>
installing gcc 4.9 on sparc
<Regenaxer>
It is not so that it is 64-bit hardware, right?
<tankfeeder>
sparc is 64bit, pil32
<Regenaxer>
I wonder if the problem in caused by some 64-bit arch problem
<Regenaxer>
pil is very sensitive to the underlying architecture
<Regenaxer>
that's why there is the distinction between pil32 and pil64
<aw->
beneroth: 18.1.10
<aw->
beneroth: 18.1.17 ;)
<beneroth>
ok
<tankfeeder>
sparc, solaris 10, gcc 4.9.2 crash
<beneroth>
aw-, we got confused by your namespace usage. So I wondered if you use the old system. but maybe we're just confused by the namespace system, not necessarily by your usage of it.
<tankfeeder>
installing gcc 5.5
<aw->
beneroth: it doesn't work in versions > 17.34
<beneroth>
aw-, especially your usage of (local arg arg arg ...) instead of (local) arg arg arg
<aw->
err.. 17.3.4
<beneroth>
aw-, oh, sounds like old namespacing to me
<aw->
beneroth: yes exactly
<beneroth>
ok
<beneroth>
thanks, so this riddle is solved :)
<aw->
beneroth: versions >= 17.3.4 should explicitly disable namespaces
<aw->
just for that lib
<Regenaxer>
Why disable?
<Regenaxer>
or, how?
<beneroth>
Regenaxer, he uses the old namespacing system. he has a environment variable that disables use of namespaces in his lib
<aw->
the environment variable `PIL_NAMESPACES=false`
<beneroth>
so it is about disabling during runtime for his lib, not doing anything in picolisp ;)
<aw->
if it's set to false, the lib doesn't use namespaces
<Regenaxer>
yes, but how to disable? Just not use it?
<Regenaxer>
ok
<beneroth>
aw-, we just comment it out. we don't make use of environment variables anywhere else yet, so this is more transparent for us.
<aw->
it doesn't disable.. it just skips (symbols) and (local) calls