Not tag 10 at the end of message

[slevin-TMX]

At address plus there is no tag 10. I think, we have to check explicitly for "10=" and check value_.begin and value_.end

[jamesdbrock]

Hi @semkin , the hffix API requires users to call message_writer::push_back_trailer to write the final CheckSum field and also calculate and write the BodyLength field value.

[slevin-TMX]

Hi @jamesdbrock
I am talking about these lines in init() in hffix.hpp:
end_.buffer_ = checksum;
end_.current_.tag_ = 10; //CheckSum
end_.current_.value_.begin_ = checksum + 3;
end_.current_.value_.end_ = checksum + 6;

Tag 10 and its content assigned without verification.
I had real-live example, when exactly after body length was not tag 10

I submitted another comment in the same day as for tag 10 and there is my proposal how to fix this and another problems

[jamesdbrock]

Yeah it's true that we don't actually check for the 10= string when reading messages. We don't need to because we already know where all the bytes should be from the BodyLength. We generally parse only what we need to. The TagValue FIX spec has a lot of extra unnecessary bytes in it.

We could do an error check for string 10= after this check for SOH.

hffix/include/hffix.hpp

Lines 2799 to 2808 in 797ce7d

	if (*(checksum - 1) != '\x01') { // check for SOH before the checksum.
	// this guarantees that at least
	// there is one SOH in the message
	// which will prevent us from
	// falling off of the end of
	// a malformed message while
	// iterating.
	invalid();
	return;
	}

And then the error check would be like

if (checksum[0] != '1' || checksum[1] != '0' || checksum[2] != '=') {
    invalid();
    return;
}