DBI preferred syntax

Hi,

dbConnect(RPostgreSQL::PostgreSQL(), "test")   #fails

Error in postgresqlNewConnection(drv, ...) :
  RS-DBI driver: (could not connect test at local on dbname "test"
)

I guess this must be something to do with the order of matching ...?

On the manual of dbConnect(drv, ...) in
http://cran.r-project.org/web/packages/DBI/DBI.pdf
the argument ... is explained as
	authorization arguments needed by the DBMS instance;
	these typically include user, password, dbname, host, port, etc.
	For details see the appropriate DBIDriver.

There is no implication that the first unnamed argument be dbname.
Because no order is specified in the manual, the parameter name should
be explicitly written always.

Yes, it is probably an accident that it worked for RSQLite and RMySQL.

Thanks,
Paul

>-- Tomoaki NISHIYAMA Advanced Science Research Center, Kanazawa 
 >University, 13-1 Takara-machi, Kanazawa, 920-0934, Japan On

>
 >

>> On Tue, Oct 14, 2014 at 5:22 PM, Paul Gilbert

>>> I reported earlier in this thread that this works:
 >>>
 >>> Type 'q()' to quit R.
 >>>> require("RPostgreSQL")
 >>> Loading required package: RPostgreSQL
 >>> Loading required package: DBI
 >>>> require("DBI")
 >>>> z <-dbConnect(RPostgreSQL::PostgreSQL(), dbname="test")   #works
 >>>
 >>> but there are some circumstance I have not isolated where if fails 
with the
 >>> same problem as this (which fails reliably):
 >>>
 >>> Type 'q()' to quit R.
 >>>> requireNamespace("RPostgreSQL")
 >>> Loading required namespace: RPostgreSQL
 >>>> require("DBI")
 >>> Loading required package: DBI
 >>>> z <- dbConnect(RPostgreSQL::PostgreSQL(), "test")
 >>> Error in postgresqlNewConnection(drv, ...) :
 >>>   RS-DBI driver: (could not connect test at local on dbname "test"
 >>> )
 >>>
 >>> For some reason, done this way it does not find my setting of 
environment
 >>> variable PGHOST, which is not localhost, and picks up test as my 
user name,
 >>> which is not correct.
 >>
 >> I suspect this is because RPostgreSQL needs some tweaks to work when
 >> it's not attached - it's possible it's using .onAttach instead of
 >> .onLoad for initialisation.
 >
 > No, it is using .onLoad and loading vs attaching does not seem to be 
the issue. I've isolated a bit better, it seems to need dbname 
indicated, which is not the case for RSQLite or RMySQL:
 >
 > dbConnect(RPostgreSQL::PostgreSQL(), dbname="test")   #works
 > dbConnect(RPostgreSQL::PostgreSQL(), "test")   #fails
 >
 >> dbConnect(RPostgreSQL::PostgreSQL(), dbname="test")   #works
 > <PostgreSQLConnection:(9341,5)>
 >> dbConnect(RPostgreSQL::PostgreSQL(), "test")   #fails
 > Error in postgresqlNewConnection(drv, ...) :
 >  RS-DBI driver: (could not connect test at local on dbname "test"
 > )
 >
 > I guess this must be something to do with the order of matching ...?
 >
 > Paul
 >
 >>
 >>
 >> Hadley
 >>