summaryrefslogtreecommitdiff
path: root/content/bridging_the_gap.md
diff options
context:
space:
mode:
Diffstat (limited to 'content/bridging_the_gap.md')
-rw-r--r--content/bridging_the_gap.md4
1 files changed, 3 insertions, 1 deletions
diff --git a/content/bridging_the_gap.md b/content/bridging_the_gap.md
index f6c98b7..735b1fd 100644
--- a/content/bridging_the_gap.md
+++ b/content/bridging_the_gap.md
@@ -83,7 +83,9 @@ However one fundamental constraint for the design of the node system is that it
Spinning up rustc every time the user adds a brush stroke is just not viable.
But if we take a look at programming languages, this is already a solved problem: Wasm also consists of a "high level" level language that has to be executed with as little of a delay as possible and also as fast as possible.
This is achieved by leveraging tiered compilation and initially just "interpreting" the binary and starting the compiler in the background to compute a more optimized binary that can be swapped out on the fly.
-We want to build something similar for graphite by using a mix of interpretation (executing instruction after instruction using a runtime to pass data from node to node)
+We want to build something similar for graphite by using a mix of interpretation (executing instruction after instruction using a runtime to pass data from node to node) and just in time compilation.
+But this also means that we have to build our own type checking.
+## Building a type system for a node based programming language