>>>>> Paul Johnson <pauljoh...@gmail.com>
>>>>> on Thu, 2 Jan 2014 22:29:29 -0600 writes:
> Happy New Year I recognize this is a low priority issue,
> but... I'll fix it if you let me.
> There are some TABs where R style calls for 4 spaces.
This is your not quite correct interpretation.
The R internal manual you mention below talks about TABs in
specific places in *.Rd files only.
*.R (and *.c even more) use TABs in many places, and all of R core have
their Emacs / vi / ... settings accordingly.
Happy New Year, indeed, to all contributors and readers of 'R-devel'!
Martin
--
Martin Maechler, ETH Zurich & R core team
> For example R-3.0.2/src/library/stats/R/dummy.coef.R.
> I never noticed this until today, when I was stranded on a
> deserted island with only the R source code and a Swiss
> Army knife (vi). Now I realize my ~.vimrc has tabstop set
> at 2, and it makes tab indentations in R source "go
> backwards". if you are strictly following R Internals 9
> "R Coding Standards", the TABs should be replaced with
> spaces, I shouldn't have to reconfigure the editor.
> Here's the way it looks in my vi
> dummy.coef.aovlist <- function(object, use.na = FALSE,
> ...) { Terms <- terms(object, specials="Error") err <-
> attr(Terms,"specials")$Error - 1 tl <- attr(Terms,
> "term.labels")[-err] int <- attr(Terms, "intercept") facs
> <- attr(Terms, "factors")[-c(1,1+err), -err, drop=FALSE]
> vars <- rownames(facs) xl <- attr(object, "xlevels")
> if(!length(xl)) { # no factors in model
> return(as.list(coef(object))) } nxl <- setNames(rep.int(1,
> length(vars)), vars) tmp <- unlist(lapply(xl, length))
> nxl[names(tmp)] <- tmp lterms <- apply(facs, 2L,
> function(x) prod(nxl[x > 0])) nl <- sum(lterms) args <-
> setNames(vector("list", length(vars)), vars) for(i in
> vars) args[[i]] <- if(nxl[[i]] == 1) rep.int(1, nl) else
> factor(rep.int(xl[[i]][1L], nl), levels = xl[[i]]) dummy
> <- do.call("data.frame", args) pos <- 0 rn <- rep.int(tl,
> lterms) rnn <- rep.int("", nl) for(j in tl) { i <-
> vars[facs[, j] > 0] ifac <- i[nxl[i] > 1] if(length(ifac)
> == 0L) { # quantitative factor rnn[pos + 1] <- j } else
> if(length(ifac) == 1L) { # main effect dummy[
> pos+1L:lterms[j], ifac ] <- xl[[ifac]] rnn[
> pos+1L:lterms[j] ] <- as.character(xl[[ifac]]) [... snip
> ....]
> The patch looks like this in my vi:
> *** dummy.coef.R 2013-03-05 17:02:23.000000000 -0600 ---
> dummy.coef-indent.R 2014-01-02 21:49:25.828163544 -0600
> ***************
> *** 28,35 **** Terms <- delete.response(Terms) vars <-
> all.vars(Terms) xl <- object$xlevels ! if(!length(xl)) {
> # no factors in model ! return(as.list(coef(object))) }
> nxl <- setNames(rep.int(1, length(vars)), vars) tmp <-
> unlist(lapply(xl, length)) ## ?? vapply(xl, length, 1L)
> --- 28,35 ---- Terms <- delete.response(Terms) vars <-
> all.vars(Terms) xl <- object$xlevels ! if(!length(xl)) {
> # no factors in model ! return(as.list(coef(object))) }
> nxl <- setNames(rep.int(1, length(vars)), vars) tmp <-
> unlist(lapply(xl, length)) ## ?? vapply(xl, length, 1L)
> ***************
> *** 38,64 **** nl <- sum(lterms) args <-
> setNames(vector("list", length(vars)), vars) for(i in
> vars) ! args[[i]] <- if(nxl[[i]] == 1) rep.int(1, nl) !
> else factor(rep.int(xl[[i]][1L], nl), levels = xl[[i]])
> dummy <- do.call("data.frame", args) pos <- 0 rn <-
> rep.int(tl, lterms) rnn <- rep.int("", nl) for(j in tl) {
> ! i <- vars[facs[, j] > 0] ! ifac <- i[nxl[i] > 1] !
> if(length(ifac) == 0L) { # quantitative factor !
> rnn[pos+1] <- j ! } else if(length(ifac) == 1L) { # main
> effect ! dummy[ pos+1L:lterms[j], ifac ] <- xl[[ifac]] !
> rnn[ pos+1L:lterms[j] ] <- as.character(xl[[ifac]]) ! }
> else { # interaction ! tmp <- expand.grid(xl[ifac]) !
> dummy[ pos+1L:lterms[j], ifac ] <- tmp ! rnn[
> pos+1L:lterms[j] ] <- ! apply(as.matrix(tmp), 1L,
> function(x) paste(x, collapse=":")) ! } ! pos <- pos +
> lterms[j] } ## some terms like poly(x,1) will give
> problems here, so allow ## NaNs and set to NA afterwards.
> --- 38,64 ---- nl <- sum(lterms) args <-
> setNames(vector("list", length(vars)), vars) for(i in
> vars) ! args[[i]] <- if(nxl[[i]] == 1) rep.int(1, nl) !
> else factor(rep.int(xl[[i]][1L], nl), levels = xl[[i]])
> dummy <- do.call("data.frame", args) pos <- 0 rn <-
> rep.int(tl, lterms) rnn <- rep.int("", nl) for(j in tl) {
> ! i <- vars[facs[, j] > 0] ! ifac <- i[nxl[i] > 1] !
> if(length(ifac) == 0L) { # quantitative factor !
> rnn[pos+1] <- j ! } else if(length(ifac) == 1L) { # main
> effect ! dummy[ pos+1L:lterms[j], ifac ] <- xl[[ifac]] !
> rnn[ pos+1L:lterms[j] ] <- as.character(xl[[ifac]]) ! }
> else { # interaction ! tmp <- expand.grid(xl[ifac]) !
> dummy[ pos+1L:lterms[j], ifac ] <- tmp ! rnn[
> pos+1L:lterms[j] ] <- ! apply(as.matrix(tmp), 1L,
> function(x) paste(x, collapse=":")) ! } ! pos <- pos +
> lterms[j] } [...snip...]
> I could make a patch that applies at the top level, maybe
> one for each library. If one of you will consider using
> it.
> Again, I recognize this is a small thing, the editor can
> be adjusted to avoid it. But it shouldn't be necessary.
> PJ
> --
> Paul E. Johnson Professor, Political Science
> Assoc. Director 1541 Lilac Lane, Room 504 Center for
> Research Methods University of Kansas University of Kansas
> http://pj.freefaculty.org http://quant.ku.edu
> [[alternative HTML version deleted]]
> ______________________________________________
> 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
