HTTP/1.1 client with connection pooling #63

Phase 6: HTTP/1.1 Client with Connection Pooling#

Implement the high-level HTTP/1.1 client in the net crate, bringing together TCP, TLS, DNS, URL, and HTTP parsing.

Requirements#

• HttpClient type:
  • get(url: &Url) -> Result<HttpResponse>
  • post(url: &Url, body: &[u8], content_type: &str) -> Result<HttpResponse>
  • request(method: Method, url: &Url, headers: &Headers, body: Option<&[u8]>) -> Result<HttpResponse>
• Scheme handling:
  • http:// — plain TCP connection
  • https:// — TLS connection with certificate verification
• Connection lifecycle:
  • DNS resolve hostname → connect TCP → (optional TLS handshake) → send request → read response
  • Respect Connection: keep-alive / Connection: close headers
• Connection pooling:
  • Pool idle connections by (host, port, is_tls) key
  • Reuse connections for subsequent requests to same origin
  • Max idle time before closing (e.g., 60 seconds)
  • Max connections per host (e.g., 6)
• Redirect following: follow 301, 302, 307, 308 redirects (configurable max redirects)
• User-Agent: we-browser/0.1

Dependencies#

• WHATWG URL parser
• TCP socket wrapper
• DNS stub resolver
• TLS 1.3 client (handshake + record layer + key schedule)
• HTTP/1.1 parser

Acceptance Criteria#

• Fetch HTTP URLs over plain TCP
• Fetch HTTPS URLs with TLS 1.3
• Connection pooling with keep-alive reuse
• Redirect following (301, 302, 307, 308)
• Proper Host header from URL
• Timeout support for connect and read
• Integration test: fetch a real HTTPS page
• Tests (15+ unit tests + integration tests)
• cargo clippy and cargo fmt clean