[datatable-help] setkey removes attribute settings

[Matthew Dowle]

Now in v1.6.2 on R-Forge :

  o  setkey no longer copies the whole table and should be
     faster for large tables. Each column is reordered by reference
     (in C) using one column of working memory, FR#1006. User
     defined attributes on the original table are now also
     retained (thanks to Thell Fowler for reporting).


On Thu, 2011-06-09 at 07:52 -0700, Thell wrote:
> On Jun 9, 2:18 am, Matthew Dowle <mdo... at mdowle.plus.com> wrote:
> > ... and I added a comment and link back to this thread, to this FR :https://r-forge.r-project.org/tracker/index.php?func=detail&aid=1006&...
> > setkey doesn't retain user attributes because it (currently) copies the
> > data.table, something we'd like to improve anyway.
> > Matthew
> >
> >
> >
> > On Wed, 2011-06-08 at 22:34 -0400, Steve Lianoglou wrote:
> > > Hi,
> >
> > > On Wed, Jun 8, 2011 at 7:01 PM, Thell Fowler <tbfowl... at gmail.com> wrote:
> >
> > > > It looks like custom attributes are lost by a DF to DT creation, and when the key info for a DT is invalidated, altered, etc...  I ran into this when trying to attach some attribute info to a table that identifies column groups so that a new global variable or passing another param to functions wouldn't be needed.  This worked nicely, until I had need to either modify some data in the table, which invalidates the key...
> >
> > > > Was this a design decision?  If so, why?
> >
> > > Try `as.data.table` for a more "delicate" way to convert a data.frame
> > > to a data.table. Your attributes will come through to the other side.
> >
> > > R> df <- data.frame( id=c(1:5), name= letters[1:5] )
> > > R> attr(df, 'testing') <- 'keep me'
> > > R> dt <- as.data.table(df)
> > > R> attr(dt, 'testing')
> > > [1] "keep me"
> >
> > > -steve
> >
> >
> 
> Thank you for the quick replies, for adding this detail to the feature
> request, and for the as.data.table hint (which I'll be testing today).
> 
> I guess for now, in instances when a function will invalidate the keys
> and return the table I'll copy out the specific attributes for column
> groups, then restore them prior to passing the table back.
> 
> Sincerely,
> Thell
> _______________________________________________
> datatable-help mailing list
> datatable-help at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/datatable-help