Summary
node:http and node:https link and run (after #1846), but ~50 tests fail with byte-level output differences: http 27, https 23. The test bodies execute; the assertions compare http/https response bytes against an expected fixture and they don't match.
Examples: test-http-agent-keep-alive-timeout-buffer.js, test-http-client-defaults.js, test-http-client-error-rawbytes.js, test-https-agent-abort-controller.js, test-https-agent-disable-session-reuse.js, test-https-agent-session-reuse.js.
Likely areas
Format/framing of http(s) responses: status-line spelling/casing, default headers (Connection/Date/Content-Length/Transfer-Encoding/Server), CRLF handling, header ordering, body framing (chunked vs content-length), TLS handshake stringification (https). Each test will pin a specific difference; clustering by which header/byte differs would yield concrete sub-fixes.
Distinct from missing methods (other http/net tickets) — these tests reach the assertion successfully, the bytes just disagree.
Context: #800 sweep, perry 0.5.1032. 27 http + 23 https in the diff bucket.
Summary
node:httpandnode:httpslink and run (after #1846), but ~50 tests fail with byte-level output differences: http 27, https 23. The test bodies execute; the assertions compare http/https response bytes against an expected fixture and they don't match.Examples:
test-http-agent-keep-alive-timeout-buffer.js,test-http-client-defaults.js,test-http-client-error-rawbytes.js,test-https-agent-abort-controller.js,test-https-agent-disable-session-reuse.js,test-https-agent-session-reuse.js.Likely areas
Format/framing of http(s) responses: status-line spelling/casing, default headers (Connection/Date/Content-Length/Transfer-Encoding/Server), CRLF handling, header ordering, body framing (chunked vs content-length), TLS handshake stringification (https). Each test will pin a specific difference; clustering by which header/byte differs would yield concrete sub-fixes.
Distinct from missing methods (other http/net tickets) — these tests reach the assertion successfully, the bytes just disagree.
Context: #800 sweep, perry 0.5.1032. 27 http + 23 https in the diff bucket.