[R] x[0]: Can '0' be made an allowed index in R?

Richard O'Keefe r@oknz @end|ng |rom gm@||@com
Wed Apr 24 02:24:35 CEST 2024 

" You could
have a global setting for whether ALL operations are  0-based or 1-based"

NO.  You could, but you shouldn't.  []IO  (index origin) in APL has
always let you do this
and it has always been a good way to break lots of stuff.

A much better approach is to think of "index-from-0" and
"index-from-1" as different
operations.  Fortran, Algol, PL/I, Pascal, and Ada let you choose your
own origin for
each subscript of each array independently.  It's sometimes nice to do
  frequency : array ('A'..'Z') of natural;
But it complicates the heck out of providing general-purpose functions
for arrays as
wholes, and that's what R is all about.

Let's take Smalltalk as an example.
  sequence at: index  "1-origin"
  sequence at: index put: newValue "1-origin"
You *could* define a new kind of sequence in Smalltalk with origin 1,
but that would stuff up *so* many operations that it would be no fun at all.
(I've done it.  Never again.)
Instead, doing
  at0: index   ^self at: index + 1
  at0: index put: newValue  ^self at: index + 1 put: newValue
means that I get to use 0 origin indexing on oodles of kinds of sequences
WITHOUT BREAKING ANYTHING that uses the existing operations.

It's the same in R.  Don't think "new kind of item which is indexed from 0".
Think "new kind of indexing operation that works on anything that uses
normal indexing".  Nothing breaks.  You get to use 0-origin on all sorts of
stuff including the results of existing functions.  And human readers KNOW
at a glance where 0-origin is used and where it isn't.

Consider, for example, the built-in function head().  If you dive into the code,
you'll find that head(x, 2) does x[seq_len(2)] which is x[1:2] which
assumes 1-origin.
If you make a new data type, head won't work on it.
If you make the index origin dynamic, head will work sometimes and
fail sometimes.
If you use a new indexing operation, head works just fine.


On Wed, 24 Apr 2024 at 03:25, <avi.e.gross using gmail.com> wrote:
>
> I think it might be fair to say that the discussion is becoming a tad wider
> than whether you want your data structures indexed starting from 0 or 1.
>
> Programming languages have added functionality to do many things on top of
> the simple concept of accessing or changing the nth element one at a time.
> If someone wants to make a parallel way to handle things, it may work for
> some uses but not others and that is fine as long as the user of the package
> knows the limits and is careful.
>
> In R, as has been pointed out, there are behaviors associated with indexing
> that are NOT the only way they could have been done. These are design
> choices made long ago.
>
> Negated numbers are allowed in some contexts and have meaning. Ranges of
> numbers can be specified. Rows and columns both must be accommodated. And
> since quite a bit of the language gets written in languages like C/C++ that
> may not anticipate such a change, it MIGHT BE that there will be errors if
> the way being zero-based is introduced does not work well.
>
> I have not looked at the packages mentioned and it may well be that they are
> very carefully crafted in a way that takes everything into account and even
> behaves well if you have operations that work jointly on both 0 and 1 based
> objects. They may even have a way to identify which way a particular object
> is numbered.
>
> As pointed, out programming languages differ. I believe JavaScript at one
> point had a data structure that was sort of like a dictionary that treated
> integer indexes as a sort of indexed array and did interesting things if
> some numbers were missing or added and sort of reindexed it when used as an
> array. Features can be very powerful and at the same time be potentially
> dangerous.
>
> But the original request was not only reasonable but also something others
> had worked on and produced packages. Chances are good that many of the
> questions people had were considered and either implemented well or
> documented as something to avoid.
>
> I would like to add that R is evolving as is the way people use it. An
> example might be how you can make and use data.frames (including new
> variants) with variable list components and do transformations. I have seen
> problems when using such features because some operations on them may not do
> what is expected. In some cases, I suspect that ideas that are accepted and
> moved back into the base after careful consideration are safer. You could
> have a global setting for whether ALL operations are  0-based or 1-based or
> have several data types to choose from that work as you want. But some of it
> is simply engrained in peoples minds as in that programmers sometimes call a
> function and index the results as in minmax(x)[2] and that digit 2 is
> hardwired in with the assumption they are getting the max value stored in
> the second position and ignoring the min value stored in the first position.
> Flipping a switch at the top of the program leaves such errors in place.
> Many packages include all kinds of shortcuts like that and a global switch
> idea may leave many dangling packages.
>
> For the many of us with experiences in many programming languages, all kinds
> of mental stumbling blocks abound unless we accept each language as a world
> of its own in which different design and implementation choices were made.
> If you take a language like Python and ask how to support 1-based objects,
> you might well come up with a rather different solution including using
> hooks such as the dunder variables  they made easily accessible, making a
> subclass, using a decorator and so on.
>
> Having said that, consider that  there are many successful examples in R
> where packages have augmented doing things in ways different than base R and
> different does not have to be better or worse but can have advantages for
> some people in some situations.
>
> An example often discussed here is the a group of packages called the
> tidyverse. Some aspects of it are more appealing to me than the parts of
> base R I might use to do the same kinds of things and some don't. Within
> reason, many parts can be mixed and matched. There do seem to be places the
> match is not so great and places where they added too much functionality
> that many users do not need and which complicated earlier ways of doing
> simpler things. And, of course, it can divide a community when others cannot
> understand your code without lots of additional education and new ways of
> thinking.
>
> A well-designed way to designate individual data structures to be 0-based
> would indeed be nice to have, and apparently it is available. I have seen
> other packages mentioned here that in which I worked a bit with the one
> making the package and saw how many hooks had to be dealt with to handle
> having multiple ways of being "missing" as is available in some programs
> outside R. Many implementations only work with known cases and can break
> when combined, perhaps with other newer changes to R or packages. But, if
> used within a carefully designed environment they may do what you need and
> preserve functionality.
>
>
>
>
> -----Original Message-----
> From: R-help <r-help-bounces using r-project.org> On Behalf Of John Fox
> Sent: Tuesday, April 23, 2024 10:24 AM
> To: Peter Dalgaard <pd.mes using cbs.dk>
> Cc: R help project <r-help using r-project.org>; Hans W <hwborchers using gmail.com>
> Subject: Re: [R] x[0]: Can '0' be made an allowed index in R?
>
> Hello Peter,
>
> Unless I too misunderstand your point, negative indices for removal do
> work with the Oarray package (though -0 doesn't work to remove the 0th
> element, since -0 == 0 -- perhaps what you meant):
>
>  > library(Oarray)
>
>  > v <- Oarray(1:10, offset=0)
>
>  > v
> [0,] [1,] [2,] [3,] [4,] [5,] [6,] [7,] [8,] [9,]
>     1    2    3    4    5    6    7    8    9   10
>
>  > dim(v)
> [1] 10
>
>  > v[-1]
> [1]  1  3  4  5  6  7  8  9 10
>
>  > v[-0]
> [1] 1
>
> Best,
>   John
>
> On 2024-04-23 9:03 a.m., Peter Dalgaard via R-help wrote:
> > Caution: External email.
> >
> >
> > Doesn't sound like you got the point. x[-1] normally removes the first
> element. With 0-based indices, this cannot work.
> >
> > - pd
> >
> >> On 22 Apr 2024, at 17:31 , Ebert,Timothy Aaron <tebert using ufl.edu> wrote:
> >>
> >> You could have negative indices. There are two ways to do this.
> >> 1) provide a large offset.
> >> Offset <- 30
> >> for (i in -29 to 120) { print(df[i+Offset])}
> >>
> >>
> >> 2) use absolute values if all indices are negative.
> >> for (i in -200 to -1) {print(df[abs(i)])}
> >>
> >> Tim
> >>
> >>
> >>
> >> -----Original Message-----
> >> From: R-help <r-help-bounces using r-project.org> On Behalf Of Peter Dalgaard
> via R-help
> >> Sent: Monday, April 22, 2024 10:36 AM
> >> To: Rolf Turner <rolfturner using posteo.net>
> >> Cc: R help project <r-help using r-project.org>; Hans W <hwborchers using gmail.com>
> >> Subject: Re: [R] x[0]: Can '0' be made an allowed index in R?
> >>
> >> [External Email]
> >>
> >> Heh. Did anyone bring up negative indices yet?
> >>
> >> -pd
> >>
> >>> On 22 Apr 2024, at 10:46 , Rolf Turner <rolfturner using posteo.net> wrote:
> >>>
> >>>
> >>> See fortunes::fortune(36).
> >>>
> >>> cheers,
> >>>
> >>> Rolf Turner
> >>>
> >>> --
> >>> Honorary Research Fellow
> >>> Department of Statistics
> >>> University of Auckland
> >>> Stats. Dep't. (secretaries) phone:
> >>>         +64-9-373-7599 ext. 89622
> >>> Home phone: +64-9-480-4619
> >>>
> >>> ______________________________________________
> >>> R-help using r-project.org mailing list -- To UNSUBSCRIBE and more, see
> >>> https://stat/
> >>> .ethz.ch%2Fmailman%2Flistinfo%2Fr-help&data=05%7C02%7Ctebert%40ufl.edu
> >>> %7C79ca6aadcaee4aa3241308dc62d986f6%7C0d4da0f84a314d76ace60a62331e1b84
> >>> %7C0%7C0%7C638493933686698527%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAw
> >>> MDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C0%7C%7C%7C&sdata=
> >>> wmv9OYcMES0nElT9OAKTdjBk%2BB55bQ7BjxOuaVVkPg4%3D&reserved=0
> >>> PLEASE do read the posting guide
> >>> http://www.r/
> >>> -project.org%2Fposting-guide.html&data=05%7C02%7Ctebert%40ufl.edu%7C79
> >>> ca6aadcaee4aa3241308dc62d986f6%7C0d4da0f84a314d76ace60a62331e1b84%7C0%
> >>> 7C0%7C638493933686711061%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiL
> >>> CJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C0%7C%7C%7C&sdata=AP78X
> >>> nfKrX6B0YVM0N76ty9v%2Fw%2BchHIytw33X7M9umE%3D&reserved=0
> >>> and provide commented, minimal, self-contained, reproducible code.
> >>
> >> --
> >> Peter Dalgaard, Professor,
> >> Center for Statistics, Copenhagen Business School Solbjerg Plads 3, 2000
> Frederiksberg, Denmark
> >> Phone: (+45)38153501
> >> Office: A 4.23
> >> Email: pd.mes using cbs.dk  Priv: PDalgd using gmail.com
> >>
> >> ______________________________________________
> >> R-help using r-project.org mailing list -- To UNSUBSCRIBE and more, see
> >> https://stat.ethz.ch/mailman/listinfo/r-help
> >> PLEASE do read the posting guide
> http://www.r-project.org/posting-guide.html
> >> and provide commented, minimal, self-contained, reproducible code.
> >>
> >> ______________________________________________
> >> R-help using r-project.org mailing list -- To UNSUBSCRIBE and more, see
> >> https://stat.ethz.ch/mailman/listinfo/r-help
> >> PLEASE do read the posting guide
> http://www.R-project.org/posting-guide.html
> >> and provide commented, minimal, self-contained, reproducible code.
> >
> > --
> > Peter Dalgaard, Professor,
> > Center for Statistics, Copenhagen Business School
> > Solbjerg Plads 3, 2000 Frederiksberg, Denmark
> > Phone: (+45)38153501
> > Office: A 4.23
> > Email: pd.mes using cbs.dk  Priv: PDalgd using gmail.com
> >
> > ______________________________________________
> > R-help using r-project.org mailing list -- To UNSUBSCRIBE and more, see
> > https://stat.ethz.ch/mailman/listinfo/r-help
> > PLEASE do read the posting guide
> http://www.R-project.org/posting-guide.html
> > and provide commented, minimal, self-contained, reproducible code.
>
> ______________________________________________
> R-help using r-project.org mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>
> ______________________________________________
> R-help using r-project.org mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.

More information about the R-help mailing list