Custom HTTP redirects (Unification) #1628
Conversation
avbelov23
force-pushed
the
avb-856
branch
5 times, most recently
from
70b8716 to
30de466
Compare
There was a problem hiding this comment.
I appreciate the effort to aggregate all the response functions - the code now looks much better! There are still places to improve and/or discuss. Also please see new comments in #1618
fw/http.c
Outdated
| } | ||
| body_len = tfw_str_to_cstr(body, body_val, BODY_BUF_LEN); | ||
| if (!body_len) | ||
| return TFW_BLOCK; |
There was a problem hiding this comment.
Actually, tfw_h1_prep_resp() -> tfw_msg_write() and tfw_h2_prep_resp() -> tfw_h2_frame_local_resp() -> tfw_h2_append_predefined_body() do allocate skb memory and copy the msg chunks into it, so we're good with many chunks at this layer, so we can just append body chunks to msg and don't need to copy them
I see your comment
Remark:
The body of the js challenge redirect can consist of chunks (tfw_http_sticky_build_redirect()), but tfw_h2_frame_local_resp() is not ready to work with such a body, therefore, to avoid problems, it was temporarily decided to expand such a body into a plain tfw string
, but tfw_h2_frame_local_resp() accepts body as TfwStr, which confuses me. Let's discuss this on the call.
There was a problem hiding this comment.
tfw_h2_append_predefined_body() for some reason expects a plain string
The function is designed for @body preallocated during
configuration processing thus no chunked body allowed, only plain TfwStr is accepted there
b26974e to
028430c
Compare
#856
Remark:
The body of the js challenge redirect can consist of chunks (
tfw_http_sticky_build_redirect()), buttfw_h2_frame_local_resp()is not ready to work with such a body, therefore, to avoid problems, it was temporarily decided to expand such a body into a plain tfw string