| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| |
|
|
| |
But crashes on unmatch sometimes in production.
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
Also refactors/separates ::nodename to be ::thisNode (this computer) and
::thisProcess (this OS process).
All tests should pass. Zygote-based processes are much faster to start
up since they don't need to recompile C code or reload Folk. They
still need to connect to Folk via WebSocket for now, and the automatic
management we'd want (bidirectional, based on both match lifetime and
process lifetime) isn't working yet.
Some of the node vs. process distinction in naming still needs to be
cleaned up, too, I think.
|
| | |
|
| |
|
|
|
|
|
| |
Should hopefully be faster bc we can cache the bytecode representation
of each code block
Only working on (some of) test/basic right now
|
| |
|
|
| |
Another 5ms performance improvement, I think?
|
| |
|
|
| |
(Its Tcl implementation was the last bottleneck)
|
| |
|
|
|
|
|
|
|
|
|
| |
Joins won't work with `Assert when` now, but they will work with
ordinary `When`.
The motivation is to make it possible to use `When when` with every
clause of a join, so every clause has to exist as a separate When in
the database. (Another point is that the old join implementation was a
significant jump in core complexity without an accompanying
performance increase, so I don't really want to keep it around.)
|
| | |
|
| | |
|
| |
|
