libera/#lisp - IRC Chatlog
Search
21:17:59
amirouche
1) I woult not support remote execution of programs, except user scripts with their own distribution channel supported by trustable social network
21:19:54
amirouche
3) I would not re-invent a network format, and instead of HTTP I would use s-expr
21:22:38
amirouche
An alternative approach to the above is to expose procedures over a network that can be inspected
22:24:12
contrapunctus
diod, amirouche: or maybe a Lisp programmer would realize that a data structure/markup is never as general as a program, and that all markup languages tend to gain PL features...and thus you'd have a web of properly sandboxed programs with a well-designed PL to begin with ¯\_(ツ)_/¯
22:25:19
pl
amirouche: Lisp-oriented protocols, at least on Symbolics Genera stack, didn't use S-expressions
22:26:15
pl
or at least, S-expressions that are textual serialization. So instead you had protocols that directly handled atoms and lists and such
22:27:52
pl
contrapunctus: Genera manuals, especially on network support, available on Bitsavers. You can also unpack genera sources using OpenGenera or physical Symbolics LispM
22:32:55
pl
so you had lowest layers (L1/L2 protocols), internetworking protocols (IP, DECnet, Chaosnet, I think CLNS was an option, etc.), more intelligent protocols providing some basic services like streams and datagrams (TCP, UDP, Chaosnet streams, etc.), and on top of that you had "building blocks" for application protocols, like ability to send Lisp-conpatible data structures - and finally on top of that they build things like NFILE etc.
22:56:56
masinter
The Medley source code is pretty good testiment that we had no idea what we were doing
0:23:17
contrapunctus
What I'm curious about is how one may retain the property of user control over content and presentation (as seen on the web, e.g. user styles, ad blocking, etc) on a platform of 'programs', i.e. one lacking the markup language/stylesheet/programming language distinction.