diff options
| author | Cliff Click <cliffc@acm.org> | 2025-08-13 16:35:26 -0700 |
|---|---|---|
| committer | Cliff Click <cliffc@acm.org> | 2025-08-13 16:35:26 -0700 |
| commit | f83e8d48bc2993ca53d2e38294f1f729c5a34943 (patch) | |
| tree | 679cc7a3a78c42b996cdd95723e9e013b0330c68 | |
| parent | a2d16249ad6efe80ec010a03e849f02ae124fc58 (diff) | |
| download | Simple-f83e8d48bc2993ca53d2e38294f1f729c5a34943.tar.gz Simple-f83e8d48bc2993ca53d2e38294f1f729c5a34943.zip | |
typo fix in reader
| -rw-r--r-- | TypeAnalysis.md | 6 |
1 files changed, 3 insertions, 3 deletions
diff --git a/TypeAnalysis.md b/TypeAnalysis.md index f2f573ee..cb283dec 100644 --- a/TypeAnalysis.md +++ b/TypeAnalysis.md @@ -16,11 +16,11 @@ In both languages, after the first round of *pessimistic* constant propagation (and extensive peephole optimizations) completes, every program point (node in the IR) has a type - and if there are no error types, then the program is type-safe. No separate typing pass is needed, and this is typically done -partially as the program parses, and immediately thereafter. Optionally, an -*optimistic* constant propagation (an extended SCCP) can be run to type more +partially as the program parses, and shortly thereafter. Optionally, an +*optimistic* constant propagation (SCCP) can be run to type more programs than the *pessimistic* approach alone. -These discussion relies on the program being in SSA form, which I assume the +This discussion relies on the program being in SSA form, which I assume the reader is familiar with, and is commonly used in *constant propagation* because it allows the algorithm to be *sparse* without effort. Also, all three compilers use the Sea-of-Nodes representation which is only relevant here |