MIME-Version: 1.0
References: <92a25cd6-5b50-3074-0da6-7394db9e0f0a@chagford.com>
In-Reply-To: <92a25cd6-5b50-3074-0da6-7394db9e0f0a@chagford.com>
From: Daniele Varrazzo <daniele.varrazzo@gmail.com>
Date: Wed, 27 May 2020 00:11:34 +1200
Message-ID: 
 <CA+mi_8YDBvXkns7odQQ5VLATyCXe2xFTw_KV5H19Q9aTyWEEog@mail.gmail.com>
Subject: Re: Minor issue
To: Frank Millman <frank@chagford.com>
Cc: psycopg@postgresql.org
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

No, we don't want to add any intelligence in trying to figure out what
is into a query. If you are comfortable that you will be using always
the same pattern for comments you can easily clean the string yourself
before passing it to psycopg.

A better approach for you I guess would be to use named placeholders,
so that an a missing placeholder wouldn't require you to change the
arguments to execute.

-- Daniele

On Tue, 26 May 2020 at 23:43, Frank Millman <frank@chagford.com> wrote:
>
> Hi all
>
> This is very minor, but I thought I would mention it.
>
> I have a function that returns a complex SQL query and a tuple of
> parameters. The query is stored inside the function as a triple-quoted
> string, and the parameters are derived depending on the input arguments.
>
> Sometimes while testing I will comment out some of the SQL using '--'.
> If those lines happen to contain a parameter placeholder ('%s') I
> expected to remove the parameter from the tuple as well.
>
> pyodbc and sqlite3 both work this way, but psycopg2 raises the exception
> 'tuple index out of range'.
>
> I can live with it, but it means that I have to adjust the parameter
> tuple differently depending on which database I am testing with.
>
> If it can be fixed, that would be nice. If it can't, no problem.
>
> Frank Millman
>
>