* Add an inbox/ outbox system to server components
* Define a data flow from Request -> computation -> Update
* Create simple handlers to call server or client code for requests
A bit of a rant: it's fucking ridiculous how much work it is to subtract
two points from each other with mint + nalgebra. I ended up adding a
complete `Vector2` implementation in this commit because I got fed up
with it.
Consider this the unofficial start of librgx (rust graphics x), which
will basically become a libgdx in Rust, and the basis for the little
game engine.
Not to mention that ggez probably won't be maintained past version 0.6.0
unless some other maintainer takes over (I guess that could always happen).
Anyway, I'm annoyed!
* restructure assets directory
* implement asset loading and dynamic conversion to sprites
* reload sprites with unique URIs to load at runtime
* provide an updated renderer API to give access to client state
* use new APIs to draw a single node frame on screen
* use colour APIs to dynamically change node frame colour
The build script doesn't pass the correct path to nixpkgs in the nix-build
invocation, which causes build failures outside a standard libkookie-booted
environment.
>From jq(1):
o --raw-output / -r:
With this option, if the filter´s result is a string then it will be
written directly to standard output rather than being formatted as a
JSON string with quotes. This can be useful for making jq filters talk to
non-JSON-based systems.
This commit reverts the following commits:
* b2681a59d5.
* a35411dc74.
As already mentioned in the commit message of a2513a, this design
decision turned out to be a bad one: instead of having the iterator
abstraction branch internally, an API is required to control the flow
of iterators externally. This way users can opt-into complexity,
instead of having to opt-out.
I opted to revert the commits, instead of trying to untangle all the
changes made in these two commits, to avoid breaking any of the code.
Signed-off-by: Katharina Fey <kookie@spacekookie.de>
Previously a Yield object was either a file dump, or a simple
enumeration on a directories children, with no associated type state
for them. This commit implements a mechanism to internally fetch this
type information, to pass out to a user via the Yield type.
This way it is much easier to figure out which entries are
directories, and which are files. Additional queries have to be
passed into the FileTree for lookups. In a way, this commit gives up
on the idea of having a nested API structure for the time being, until
constructing sub-FileTrees becomes a real necessity