Re: BUG #5304: psql using conninfo fails in connecting to the server

On 02/02/2010 06:40 PM, Tom Lane wrote:
> The difference with PQconnectdbParams is that the dbname might not be
> the first thing in the parameter array. I think that a straightforward
> implementation would have the effect of the expanded dbname overriding
> parameters given before it, but not those given after it. Consider
>
> keyword[0] = "port";
> values[0] = "5678";
> keyword[1] = "dbname";
> values[1] = "dbname = db user = foo port = 9999";
> keyword[2] = "user";
> values[2] = "uu";
>
> What I'm imagining is that this would end up equivalent to
> dbname = db user = uu port = 9999. That's probably reasonable,
> and maybe even useful, as long as it's documented.

No doc changes yet, and I still have not corrected the earlier mentioned
issue,

> While I'm looking at this, there's another bogosity in the original
> patch: it neglected the PQsetdbLogin call in psql/command.c, meaning
> that doing \c would result in losing the application_name setting.

but I wanted to get feedback before going further.

This patch implements Tom's idea above. Note that I also rearranged the
parameters for the call from psql so that dbname is last, therefore
allowing a conninfo to override all other settings. The result looks like:

keywords[0] = "host";
values[0] = options.host;
keywords[1] = "port";
values[1] = options.port;
keywords[2] = "user";
values[2] = options.username;
keywords[3] = "password";
values[3] = password;
keywords[4] = "application_name";
values[4] = pset.progname;
keywords[5] = "dbname";
values[5] = (options.action == ACT_LIST_DB &&
options.dbname == NULL) ?
"postgres" : options.dbname;
keywords[6] = NULL;
values[6] = NULL;

Which produces:

# psql -U postgres "dbname=regression user=fred application_name='joe\'s
app'"
psql (8.5devel)
Type "help" for help.

regression=> select backend_start, application_name from pg_stat_activity ;
backend_start | application_name
-------------------------------+------------------
2010-02-02 21:44:55.969202-08 | joe's app
(1 row)

regression=> select current_user;
current_user
--------------
fred
(1 row)

Are there any of the psql parameters that we do not want to allow to be
overridden by the conninfo string?

Joe