Re: Ada DB bindings and APQ

["Warren W. Gay VE3WWG" <ve3wwg@NoSpam.cogeco.ca>]

Brian May wrote:
>>>>>>"Warren" == Warren W Gay VE3WWG <ve3wwg@NoSpam.cogeco.ca> writes:
> 
> 
>     Warren> I like the neatness of this approach, but the danger is
>     Warren> that you might do the "..." part on a row that might
>     Warren> not exist. Though I suppose the Query_Type object could
>     Warren> maintain enough state such that it can raise an exception
>     Warren> if any "Values" were fetched from the row that is not there.
>     Warren> Just a quibble: I'd prefer something like "No_Row(Q)" perhaps.
> 
>     Warren> An even neater loop is this, IMHO ;-)
> 
>     >> loop
>     >> Fetch(Q);
>     >> exit when No_More_Data(Q);
>     >> 
>     >> ...
>     >> end loop;
> 
> How about
> 
> Valid_Row : Boolean;
> 
> loop
>         Fetch(Q, Valid_Row);
>         exit when not Valid_Row;
>         ...
> end loop;
> 
> Where Valid_Row is an out variable from Fetch.
> 
> This would avoid breaking any existing software, and make it obvious
> that Valid_Row has to be checked.

I like that idea.

> Other questions that might be significant, especially to some of my
> proposed changes.
> 
> 1. What happens if an unexpected exception occurs within the above
>    loop, and it prevents continuing the loop. In fact, the code that
>    handles the exception may not realize it occurred in middle of the
>    fetch operation. Is there any way of recovering the database
>    connection?

An unexpected error would raise SQL_Error (or somesuch). So
if you expect to recover from it, you must obviously be willing
to catch the exception at some level.

> 2. With Mysql what happens if I do this:
> 
> declare
>         Q1 : Root_Query_Type'Class := New_Query(C);
>         Q2 : Root_Query_Type'Class := New_Query(C);
> begin
>         Prepare(Q1, ...);
>         Prepare(Q2, ...)
> 
>         Set_Fetch_Mode(Q1, Sequential_Fetch);
>         Set_Fetch_Mode(Q2, Sequential_Fetch);
> 
>         Execute(Q1, C);
>         Execute(Q2, C);
> 
>         loop
>                 Fetch(Q1);
>                 Fetch(Q2);
>                 ...
>         end loop;
> end;
> 
> Will this work, trigger an error, or will it kill the Mysql connection
> as requests are intermixed?

I haven't tried this, but I know that PostgreSQL is supposed
to support this, and I believe it will. Whether MySQL and Sybase
can, I am not sure.

In MySQL's case, what I believe happens is that any Q2 rows that
are skipped over to fetch rows for Q1, will just be loaded
into the C client memory for later use. But I have not taken
the time to test this.
--
Warren W. Gay VE3WWG
http://home.cogeco.ca/~ve3wwg