Closed hikaricai closed 2 years ago
Anyone?
@hikaricai it would probably be easier to review this if we had a test that failed without this change.
At a quick glance, this is going to incur an additional (unnecessary?) allocation on each call.
@olix0r thanks, I will write a test case.
It seems that the "fn consum" is used only if the decoding is success, but the "fn decode_string" may only be part of decoding. For example, decoding both header name and header value.
I met decoding failure when receiving continuation header which just spilt header name and header value.
I have just updated the commit, could anyone helps to review?
Not sure if @carllerche has bandwidth to review, but I think he knows this code best. Also I think @seanmonstar and @nox have worked on this module ~recently. If no one's available, I'll try to take a look at this but it will take me some time to load up context.
Could anyone helps to review? I need this patch to fix bug in my project.
Sorry about that, I'll take a look now.
@seanmonstar Is the new commit clearly enough?
The error situation is shown in the new test case
Decoding error when processing continuation frames