Prof Brian Ripley <[EMAIL PROTECTED]> writes: > Those functions are messy (StringFromRaw was always private), and are > really just part of coerceVector and asInteger etc (which are the public > interface). > > What RdbiPgSql does is not currently correct. It has > > PROTECT(rowNames = allocVector(STRSXP, rows)); > for (row = 0; row < rows; ++row) > SET_VECTOR_ELT(rowNames, row, StringFromInteger(row+1, NULL)); > > but you should not be applying SET_VECTOR_ELT to a STRSXP. The intended > effect is obtained by > > PROTECT(rowNum = allocVector(INTSXP, rows)); > irow = INTEGER(rowNum); > for(row = 0; row < rows; row++) irow[row] = row+1; > PROTECT(rowNames = coerceVector(rowNames, STRSXP)); > > (in R terms, as.character(1:row)). However, in R-devel, you can make use > of the space-optimized versions of row.names, so you don't need the > coercion and would be better off without it. > > Finally, this is done unconditionally, but only used if matrix == > FALSE.
Thanks for the clarification and example. I will make sure RdbiPgSql gets fixed. + seth ______________________________________________ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
