You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
* robust EoL processing via real FSM states w/o ugly _eol;
* remove pseudo states like RGEN_LWS_empty and process empty Host in exactly
this header processing routine;
* do not play dirty games with parser->hdr fixup which could be s_line or whatever,
but rather explicitly move w/o fixups (probably, #487 is obsolete now);
* now hm->parser.to_read = 0 means 'everything is read' whic is different from
-1, 'do not know how much to read';
* merge tfw_http_msg_field_chunk_fixup() with __tfw_http_msg_add_data_ptr();
* add new unit tests for HTTP parser;
* there is no LWS in RFC 7230, but rather OWS;
* many code cleanups.
@aleksostapenko could you please revise the task with HTTP/2 parser changes? It should be quickly to do with the HTTP/2 changes or we could just close it.
It seems the only sense of the change is to point &msg->conn->parser.hdr to status line and use generic macros, e.g. __FSM_MATCH_MOVE() and __FSM_MOVE_n() instead of field parametrized __FSM_MATCH_MOVE_f() and __FSM_MOVE_nf(). HTTP/2 treats status line in the same HEADER frame and according the same rules.
It likely will make the code easier by unifying EOL-based header line parsing.
The text was updated successfully, but these errors were encountered: