Re: binary protocol, again

On Fri, Jul 20, 2012 at 6:34 PM, P. Christeas <xrg@linux.gr> wrote:
> On Friday 20 July 2012, you wrote:
>> On Fri, Jul 20, 2012 at 6:09 PM, P. Christeas <xrg@linux.gr> wrote:
>> >> 3. using PQexecParams instead PQexec
>> >
>> > Yes, when query is eligible.
>>
>> Uhm... what is an eligible query?
> One that is single (not multi with semicolons), and NOT a "COMMENT" or
> "EXECUTE" one.

You should be hunting for semicolons in the query with knowledge of
comments (single line, multiline, nested), strings (including
dollar-quoted strings, unicode-escape, standard-conforming)... It
seems a painful road.

>> Are you changing the behaviour of the current cursor class or
>> introducing a new one? I'd go most definitely for the second, as I'm
>> afraid the first would break a lot of the existing usage.
>  Introducing a new one "cursor_bin", as we've discussed before.

Great :) Sorry though, I haven't got it: have you decided introducing
now the cursor_bin, hence you don't need parsing the query to see if
it's eligible or were you already using the cursor_bin? In the latter
case, why are you parsing the query and not always using PQexecParams?

>> In particular, what I'd like to see is a protocol ISQLParam, to live
>> together with the current ISQLQuote, in order to create a clear
>> distinction between the capabilities of the two cursors and avoid
>> bloating the current adapters interface.
>
> Hmm. I named it "BSQLQuote", but I guess "ISQLParam is a much better name.
> I'll change that.

Lovely (not bikeshedding about the name: the important is the
protocols to be distinct)


Thank you again

-- Daniele