Proposed patch (I think .txt files work OK as attachments to the list?)
On 2019-04-04 2:21 a.m., Martin Maechler wrote:
>>>>>> Ben Bolker
>>>>>> on Fri, 29 Mar 2019 12:34:50 -0400 writes:
>
> > I suspect that the issue is addressed (obliquely) in the examples,
> > which shows that variables with spaces in them (or otherwise
> > 'non-syntactic', i.e. not satisfying the constraints of legal R symbols)
> > can be handled by protecting them with backticks (``)
>
> > ## using non-syntactic names:
> > reformulate(c("`P/E`", "`% Growth`"), response = as.name("+-"))
>
> > It seems to me there could be room for a *documentation* patch (stating
> > explicitly that if termlabels has length > 1 its elements are
> > concatenated with "+", and explicitly stating that non-syntactic names
> > must be protected with back-ticks). (There is a little bit of obscurity
> > in the fact that the elements of termlabels don't have to be
> > syntactically valid names: many will be included in formulas if they can
> > be interpreted as *parseable* expressions, e.g. reformulate("x<2"))
>
> > I would be happy to give it a shot if the consensus is that it would
> > be worthwhile.
>
> I think it would be worthwhile to add to the docs a bit.
>
> [With currently just your and my vote, we have a 100% consensus
> ;-)]
>
> Martin
>
> > One workaround to the OP's problem is below (may be worth including
> > as an example in docs)
>
> >> z <- c("a variable","another variable")
> >> reformulate(z)
> > Error in parse(text = termtext, keep.source = FALSE) :
> > <text>:1:6: unexpected symbol
> > 1: ~ a variable
> > ^
> >> reformulate(sprintf("`%s`",z))
> > ~`a variable` + `another variable`
>
>
>
>
> > On 2019-03-29 11:54 a.m., J C Nash wrote:
> >> The main thing is to post the "small reproducible example".
> >>
> >> My (rather long term experience) can be written
> >>
> >> if (exists("reproducible example") ) {
> >> DeveloperFixHappens()
> >> } else {
> >> NULL
> >> }
> >>
> >> JN
> >>
> >> On 2019-03-29 11:38 a.m., Saren Tasciyan wrote:
> >>> Well, first I can't sign in bugzilla myself, that is why I wrote here
> first. Also, I don't know if I have the time at
> >>> the moment to provide tests, multiple examples or more. If that is
> not ok or welcomed, that is fine, I can come back,
> >>> whenever I have more time to properly report the bug.
> >>>
> >>> I didn't find the existing bug report, sorry for that.
> >>>
> >>> Yes, it is related. My problem was that I have column names with
> spaces and current solution doesn't solve it. I have a
> >>> solution, which works for me and maybe also for others.
> >>>
> >>> Either, someone can register me to bugzilla or I can post it here,
> which could give some direction to developers. I
> >>> don't mind whichever is preferred here.
> >>>
> >>> Best,
> >>>
> >>> Saren
> >>>
> >>>
> >>> On 29.03.19 09:29, Martin Maechler wrote:
> >>>>>>>>> Saren Tasciyan
> >>>>>>>>> on Thu, 28 Mar 2019 17:02:10 +0100 writes:
> >>>> > Hi,
> >>>> > I have found a bug in reformulate function and have a
> solution for it. I
> >>>> > was wondering, where I can submit it?
> >>>>
> >>>> > Best,
> >>>> > Saren
> >>>>
> >>>>
> >>>> Well, you could have given a small reproducible example
> >>>> depicting the bug, notably when posting here:
> >>>> Just a prose text with no R code or other technical content is
> >>>> almost always not really appropriate fo the R-devel mailing list.
> >>>>
> >>>> Further, in such a case you should google a bit and hopefully
> >>>> have found
> >>>> https://www.r-project.org/bugs.html
> >>>>
> >>>> which also mention reproducibility (and many more useful things).
> >>>>
> >>>> Then it also tells you about R's bug repository, also called
> >>>> "R's bugzilla" at https://bugs.r-project.org/
> >>>>
> >>>> and if you are diligent (but here, I'd say bugzilla is
> >>>> (configured?) far from ideal), you'd also find bug PR#17359
> >>>>
> >>>> https://bugs.r-project.org/bugzilla/show_bug.cgi?id=17359
> >>>>
> >>>> which was reported already on Nov 2017 .. and only fixed
> >>>> yesterday (in the "cleanup old bugs" process that happens
> >>>> often before the big new spring release of R).
> >>>>
> >>>> So is your bug the same as that one?
> >>>>
> >>>> Martin
> >>>>
> >>>> > --
> >>>> > Saren Tasciyan
> >>>> > /PhD Student / Sixt Group/
> >>>> > Institute of Science and Technology Austria
> >>>> > Am Campus 1
> >>>> > 3400 Klosterneuburg, Austria
> >>>>
> >>>> > ______________________________________________
> >>>> > R-devel@r-project.org mailing list
> >>>> > https://stat.ethz.ch/mailman/listinfo/r-devel
> >>>>
> >>>> ______________________________________________
> >>>> R-devel@r-project.org mailing list
> >>>> https://stat.ethz.ch/mailman/listinfo/r-devel
> >>
> >> ______________________________________________
> >> R-devel@r-project.org mailing list
> >> https://stat.ethz.ch/mailman/listinfo/r-devel
> >>
>
> > ______________________________________________
> > R-devel@r-project.org mailing list
> > https://stat.ethz.ch/mailman/listinfo/r-devel
>
Index: delete.response.Rd
===================================================================
--- delete.response.Rd (revision 76312)
+++ delete.response.Rd (working copy)
@@ -37,7 +37,11 @@
model. There is also a \code{"[.terms"} method to perform the same
function (with \code{keep.response = TRUE}).
- \code{reformulate} creates a formula from a character vector.
+ \code{reformulate} creates a formula from a character vector. If
+ \code{length(termlabels)}>1, its elements are concatenated with \code{+}.
+ Non-syntactic names (e.g. containing spaces or special characters; see
+ \code{\link{make.names}}) must be protected with backticks (see examples).
+
}
\value{
\code{delete.response} and \code{drop.terms} return a \code{terms}
@@ -65,6 +69,9 @@
## using non-syntactic names:
reformulate(c("`P/E`", "`\% Growth`"), response = as.name("+-"))
+x <- c("a name","another name")
+reformulate(sprintf("`%s`",x))
+
stopifnot(identical( ~ var, reformulate("var")),
identical(~ a + b + c, reformulate(letters[1:3])),
identical( y ~ a + b, reformulate(letters[1:2], "y"))
______________________________________________
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
