setcolorder could accept a vector of column names of length lower than the number of columns

[matthieugomez]

It would be nice if setcolorder could accept argument of length lower than the width of the datatable, so that

setcolorder(DT, "v3")

could by typed instead of

setcolorder( DT, c("v3", names(DT[names(DT) != "v3"])) )

As an outsider point of view, it is also more consistent with the fact that setnames can be used with one name instead of the whole list of names.

[nigmastar]

I think that he wants to move col V3 to the first position. It would be
better this way maybe:

setcolorder(df, "V3", 1)

With a third optional parameter (that must have same length of the second
and have no duplicates) giving the position(s) of the column(s)
On 3 Sep 2014 19:41, "eantonya" notifications@github.com wrote:

Can you please explain what exactly your first command is supposed to do?

—
Reply to this email directly or view it on GitHub
#798 (comment)
.

[eantonya]

@nigmastar that syntax makes a lot more sense to me than the OP one. Perhaps one can do

setcolorder(df, c('V4', 'V3'), c(1,4))

as a multiple column generalization of your proposal.

[matthieugomez]

The third argument would make it even better! I'd still like a default to seq(1:length_of_secon_argument) though - in particular for cases when the length is not a priori known

   setcolorder(DT,names(DT)[str_detect(names(DT),"id" )])

I personally reorder columns when I want specific columns to be in the first positions - for instance, I want to see them more clearly in head() or in some viewer.

[arunsrinivasan]

Possible duplicate of #592.

[MichaelChirico]

In fact my PR #1792 fixes #592 by implementing this issue. Just noticed this...

[mattdowle]

Yes, solved too thanks to @MichaelChirico 's PR #1762 now merged in v1.10.5.