A regression was introduced in this bit of the SWAR changeset which manifests as incorrectly classifying a read() with the contents GET as a token error (because there's no characters yet to parse for the path part) rather than as an Ok(Status::Partial).
This diff is three independent commits to:
Add specific regression tests covering this and similar situations
Add a general regression test checking all permutations of a partial one-line request
Fix the behavior of the parse_uri() function to disambiguate between an empty path followed by a space, versus a partial read that ends at the start of the path.
NB. This'll conflict with #175 — not in a material way, they'll work fine together, just a note.
A regression was introduced in this bit of the SWAR changeset which manifests as incorrectly classifying a
read()
with the contentsGET
as a token error (because there's no characters yet to parse for the path part) rather than as anOk(Status::Partial)
.This diff is three independent commits to:
parse_uri()
function to disambiguate between an empty path followed by a space, versus a partial read that ends at the start of the path.NB. This'll conflict with #175 — not in a material way, they'll work fine together, just a note.