RE: libpq debug log - Mailing list pgsql-hackers

From: Alvaro Herrera <alvherre@alvh.no-ip.org>
> In pqTraceOutputString(), you can use the return value from fprintf to
> move the cursor -- no need to count chars.

Yes, precisely, 2 bytes for the double quotes needs to be subtracted as follows:

    len = fprintf(...);
    *cursor += (len - 2);


> I still think that the message-type specific functions should print the
> message type, rather than having the string arrays.

I sort of think so to remove the big arrays.  But to minimize duplicate code, I think the code structure will look
like:

    fprintf(timestamp, length);
    switch (type)
    {
        case '?':
            pqTraceOutput?(...);
        break;
        case '?':
            /* No message content */
            fprintf("message_type_name");
        break;
    }    

void
pqTraceOutput?(...)
{
    fprintf("message_type_name");
    print message content;    
}

The order of length and message type is reversed.  The .sgml should also be changed accordingly.  What do you think?

Iwata-san, 
Why don't you submit v27 patch with the current arrays kept, and then v28 with the arrays removed soon after?


From: Kyotaro Horiguchi <horikyota.ntt@gmail.com>
> It would help when the value is "255", which is confusing between -1
> (or 255) in byte or 255 in 2-byte word. Or when the value looks like
> broken. I'd suggest "b"yte for 1 byte, "s"hort for 2 bytes, "l"ong for
> 4 bytes ('l' is confusing with '1', but anyway it is not needed)..

I don't have a strong opinion on this.  (I kind of think I want to see unprefixed numbers; readers will look at the
protocolreference anyway.)  I'd like to leave this up to Iwata-san and Alvaro-san.
 


    Regards
Takayuki Tsunakawa}