[R-sig-DB] DBI column names

Gabor Grothendieck ggrothend|eck @end|ng |rom gm@||@com
Tue Jul 24 02:24:37 CEST 2007


Yes.  One idea for the name for the argument might be
check.names= . This is the same argument name used
in read.table.  The value could be TRUE, FALSE or a custom
function.  TRUE would use make.db.names and FALSE would
use the identity function.  The default would be TRUE for
compatibility with the existing function.

Another possibility is associating the value with the driver
or the connection; however, that may seem inconsistent to
users if dbWriteTable uses it and other functions do not.

On 7/23/07, Seth Falcon <sfalcon using fhcrc.org> wrote:
> >> Gabor Grothendieck wrote:
> >> > It seems that the DBI package changes column names that have
> >> > dots in them replacing the dots with underscores.  Is this really
> >> > necessary?
>
> The DBI package provides a generic make.db.names and a deafult
> method.  In RSQLite, I believe this is only called when tables are
> created using dbWriteTable.  So it isn't as if DBI (or RSQLite) is
> preventing you from using (strange) column names in your DBs. -- Just
> making it a bit difficult ;-)
>
> The problem with changing this function is that anyone relying on
> (coding around) this feature is going to get bitten.  Maybe we want an
> additional argument to the dbWriteTable method?
>
> + seth
>
> --
> Seth Falcon | Computational Biology | Fred Hutchinson Cancer Research Center
> http://bioconductor.org
>




More information about the R-sig-DB mailing list