Re: [Rd] R-devel does not update the C++ returned variables
> Hervé Pagès
> on Mon, 2 Mar 2015 13:00:47 -0800 writes:
> Hi,
> On 03/02/2015 12:18 PM, Dénes Tóth wrote:
>>
>>
>> On 03/02/2015 04:37 PM, Martin Maechler wrote:
>>>
On 2 March 2015 at 09:09, Duncan Murdoch wrote:
| I generally recommend that people use Rcpp, which hides a lot of the
| details. It will generate your .Call calls for you, and generate the
| C++ code that receives them; you just need to think about the real
| problem, not the interface. It has its own learning curve, but I
think
| it is easier than using the low-level code that you need to work
with .Call.
>>>
Thanks for that vote, and I second that.
>>>
And these days the learning is a lot flatter than it was a decade ago:
>>>
R> Rcpp::cppFunction("NumericVector doubleThis(NumericVector x) {
return(2*x); }")
R> doubleThis(c(1,2,3,21,-4))
[1] 2 4 6 42 -8
R>
>>>
That defined, compiled, loaded and run/illustrated a simple function.
>>>
Dirk
>>>
>>> Indeed impressive, ... and it also works with integer vectors
>>> something also not 100% trivial when working with compiled code.
>>>
>>> When testing that, I've went a step further:
>>>
>>> ## now "test":
>>> require(microbenchmark)
>>> i <- 1:10
>>
>> Note that the relative speed of the algorithms also depends on the size
>> of the input vector. i + i becomes the winner for longer vectors (e.g. i
>> <- 1:1e6), but a proper Rcpp version is still approximately twice as
fast.
> The difference in speed is probably due to the fact that R does safe
> arithmetic. C or C++ do not:
>> doubleThisInt(i)
> [1] 2147483642 2147483644 2147483646 NA -2147483646
> -2147483644
>> 2L * i
> [1] 2147483642 2147483644 2147483646 NA NA NA
> Warning message:
> In 2L * i : NAs produced by integer overflow
> H.
Exactly, excellent, Hervé!
Luke also told me so in a private message.
and 'i+i' is looking up 'i' twice which is relatively costly
for very small i as in my example.
This ("no safe integer arithmetic in C, but in R") is another
good example {as Martin Morgan's} why using
Rccp -- or .Call() directly -- may be a too sharp edged sword and
maybe should be advocated for good programmers only.
Martin
>>
>> Rcpp::cppFunction("NumericVector doubleThisNum(NumericVector x) {
>> return(2*x); }")
>> Rcpp::cppFunction("IntegerVector doubleThisInt(IntegerVector x) {
>> return(2*x); }")
>> i <- 1:1e6
>> mb <- microbenchmark::microbenchmark(doubleThisNum(i), doubleThisInt(i),
>> i*2, 2*i, i*2L, 2L*i, i+i, times=100)
>> plot(mb, log="y", notch=TRUE)
>>
>>
>>> (mb <- microbenchmark(doubleThis(i), i*2, 2*i, i*2L, 2L*i, i+i,
>>> times=2^12))
>>> ## Lynne (i7; FC 20), R Under development ... (2015-03-02 r67924):
>>> ## Unit: nanoseconds
>>> ## expr min lq mean median uq max neval cld
>>> ## doubleThis(i) 762 985 1319.5974 1124 1338 17831 4096 b
>>> ## i * 2 124 151 258.4419164 221 4 4096 a
>>> ## 2 * i 127 154 266.4707169 216 20213 4096 a
>>> ## i * 2L 143 164 250.6057181 234 16863 4096 a
>>> ## 2L * i 144 177 269.5015193 237 16119 4096 a
>>> ## i + i 152 183 272.6179199 243 10434 4096 a
>>>
>>> plot(mb, log="y", notch=TRUE)
>>> ## hmm, looks like even the simple arithm. differ slightly ...
>>> ##
>>> ## ==> zoom in:
>>> plot(mb, log="y", notch=TRUE, ylim = c(150,300))
>>>
>>> dev.copy(png, file="mbenchm-doubling.png")
>>> dev.off() # [ <- why do I need this here for png ??? ]
>>> ##--> see the appended *png graphic
>>>
>>> Those who've learnt EDA or otherwise about boxplot notches, will
>>> know that they provide somewhat informal but robust pairwise tests on
>>> approximate 5% level.
>>> From these, one *could* - possibly wrongly - conclude that
>>> 'i * 2' is significantly faster than both 'i * 2L' and also
>>> 'i + i' which I find astonishing, given that i is integer here...
>>>
>>> Probably no reason for deep thoughts here, but if someone is
>>> enticed, this maybe slightly interesting to read.
>>>
>>> Martin Maechler, ETH Zurich
>>>
>>>
>>>
>>> __
>>> 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
> --
> Hervé Pagès
> Program in Computational Biology
> Division of Public Health Scien
Re: [Rd] Import data set from another package?
On 02/03/2015 22:48, Therneau, Terry M., Ph.D. wrote:
I've moved nlme from Depends to Imports in my coxme package. However, a
few of the examples for lmekin use one of the data sets from nlme. This
is on purpose, to show how the results are the same and how they differ.
If I use data(nlme::ergoStool) the data is not found,
data(nlme:::ergoStool) does no better.
If I add importFrom(nlme, "ergoStool") the error message is that
ergoStool is not exported.
There likely is a simple way, but I currently don't see it.
There were some off-the-mark suggestions in this thread. If you just
want a dataset from a package, use
data("ergoStool", package = "nlme")
In particular, it is somewhat wasteful to load a large namespace like
nlme when it is not needed.
--
Brian D. Ripley, rip...@stats.ox.ac.uk
Emeritus Professor of Applied Statistics, University of Oxford
1 South Parks Road, Oxford OX1 3TG, UK
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] [R] Why does R replace all row values with NAs
Diverted from R-help :
as it gets into musing about new R language "primitives"
> William Dunlap
> on Fri, 27 Feb 2015 08:04:36 -0800 writes:
> You could define functions like
> is.true <- function(x) !is.na(x) & x
> is.false <- function(x) !is.na(x) & !x
> and use them in your selections. E.g.,
>> x <- data.frame(a=1:10,b=2:11,c=c(1,NA,3,NA,5,NA,7,NA,NA,10))
>> x[is.true(x$c >= 6), ]
> a b c
> 7 7 8 7
> 10 10 11 10
> Bill Dunlap
> TIBCO Software
> wdunlap tibco.com
Yes; the Matrix package has had these
is0 <- function(x) !is.na(x) & x == 0
isN0 <- function(x) is.na(x) | x != 0
is1 <- function(x) !is.na(x) & x # also == "isTRUE componentwise"
namespace hidden for a while [note the comment of the last one!]
and using them for readibility in its own code.
Maybe we should (again) consider providing some versions of
these with R ?
The Matrix package also has had fast
allFalse <- all0 <- function(x) .Call(R_all0, x)
anyFalse <- any0 <- function(x) .Call(R_any0, x)
##
## anyFalse <- function(x) isTRUE(any(!x)) ## ~= any0
## any0 <- function(x) isTRUE(any(x == 0))## ~= anyFalse
namespace hidden as well, already, which probably could also be
brought to base R.
One big reason to *not* go there (to internal C code) at all with R is that
S3 and S4 dispatch for '==' ('!=', etc, the 'Compare' group generics)
and 'is.na() have been known and package writers have
programmed methods for these.
To ensure that S3 and S4 dispatch works "correctly" also inside
such new internals is much less easily achieved, and so
such a C-based internal function is0() would no longer be
equivalent with!is.na(x) & x == 0
as soon as 'x' is an "object" with a '==', 'Compare' and/or an is.na() method.
OTOH, simple R versions such as your 'is.true', called 'is1'
inside Matrix maybe optimizable a bit by the byte compiler (and
jit and other such tricks) and still keep the full
semantic including correct method dispatch.
Martin Maechler, ETH Zurich
> On Fri, Feb 27, 2015 at 7:27 AM, Dimitri Liakhovitski <
> dimitri.liakhovit...@gmail.com> wrote:
>> Thank you very much, Duncan.
>> All this being said:
>>
>> What would you say is the most elegant and most safe way to solve such
>> a seemingly simple task?
>>
>> Thank you!
>>
>> On Fri, Feb 27, 2015 at 10:02 AM, Duncan Murdoch
>> wrote:
>> > On 27/02/2015 9:49 AM, Dimitri Liakhovitski wrote:
>> >> So, Duncan, do I understand you correctly:
>> >>
>> >> When I use x$x<6, R doesn't know if it's TRUE or FALSE, so it returns
>> >> a logical value of NA.
>> >
>> > Yes, when x$x is NA. (Though I think you meant x$c.)
>> >
>> >> When this logical value is applied to a row, the R says: hell, I don't
>> >> know if I should keep it or not, so, just in case, I am going to keep
>> >> it, but I'll replace all the values in this row with NAs?
>> >
>> > Yes. Indexing with a logical NA is probably a mistake, and this is one
>> > way to signal it without actually triggering a warning or error.
>> >
>> > BTW, I should have mentioned that the example where you indexed using
>> > -which(x$c>=6) is a bad idea: if none of the entries were 6 or more,
>> > this would be indexing with an empty vector, and you'd get nothing, not
>> > everything.
>> >
>> > Duncan Murdoch
>> >
>> >
>> >>
>> >> On Fri, Feb 27, 2015 at 9:13 AM, Duncan Murdoch
>> >> wrote:
>> >>> On 27/02/2015 9:04 AM, Dimitri Liakhovitski wrote:
>> I know how to get the output I need, but I would benefit from an
>> explanation why R behaves the way it does.
>>
>> # I have a data frame x:
>> x = data.frame(a=1:10,b=2:11,c=c(1,NA,3,NA,5,NA,7,NA,NA,10))
>> x
>> # I want to toss rows in x that contain values >=6. But I don't want
>> to toss my NAs there.
>>
>> subset(x,c<6) # Works correctly, but removes NAs in c, understand
why
>> x[which(x$c<6),] # Works correctly, but removes NAs in c, understand
>> why
>> x[-which(x$c>=6),] # output I need
>>
>> # Here is my question: why does the following line replace the
values
>> of all rows that contain an NA # in x$c with NAs?
>>
>> x[x$c<6,] # Leaves rows with c=NA, but makes the whole row an NA.
>> Why???
>> x[(x$c<6) | is.na(x$c),] # output I need - I have to be
>> super-explicit
>>
>> Thank you very much!
>> >>>
>> >>> Most of your examples (except the ones using which()) are doing
logical
>> >>> indexing. In logical indexing, TRUE keeps a line, FALSE drops the
>> line,
>> >>> and NA returns NA. Since "x$c < 6" is NA if x$c is NA, you get the
>> >>> third kind of indexing.
>> >>>
>> >>> Your last example
[Rd] Asking for tasks of summer code 2015
Hey everyone: I am a Master student from Saarland Unirversity, Germany with the major of Bioinformatics. And I am interested in statistical learning which is also my major work in the future with the implementation by R. So I 'd like join the google summer code this year by doing tasks in your community. However I can not find whether there are tasks available provided for this year, anyone can tell me? Hank Cao [[alternative HTML version deleted]] __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
[Rd] Asssistance
Hi to All,
I am building a package in R and whenever I run command "R CMD build OAR"
in the terminal, I get the following error:
* checking for file ‘OAR/DESCRIPTION’ ... OK
* preparing ‘OAR’:
* checking DESCRIPTION meta-information ... ERROR
Malformed Depends or Suggests or Imports or Enhances field.
Offending entries:
R (>=3.0.2)
Entries must be names of packages optionally followed by '<=' or '>=',
white space, and a valid version number in parentheses.
See the information on DESCRIPTION files in section 'Creating R
packages' of the 'Writing R Extensions' manual.
This is my first time to build a package using R and it's very hard for me
to figure out where the problem is. I kindly call for your assistance in
fixing the problem. Below is my function;
bcidata <- read.csv("~/Desktop/Files_for_Package/data.csv"); bcidata
Modelsfunc<- function(bcidata){
occupancymean.data.frame <- NULL
for (k in seq(2.5,250,by=2.5)){
i <- 1000/k
j <- 500/k
bcidata$Xgrid <- cut(bcidata$PX, breaks = i, include.lowest = T)
bcidata$Ygrid <- cut(bcidata$PY, breaks = j, include.lowest = T)
bcidata$IDgrid <- with(bcidata, interaction(Xgrid,Ygrid))
bcidata$IDNgrid <- factor(bcidata$IDgrid)
levels(bcidata$IDgrid) <- seq_along(levels(bcidata$IDgrid))
bcidata$count <- ave(bcidata$PX, bcidata$IDgrid, FUN = length)
aggregate <- aggregate(bcidata$PX,bcidata[,c("Xgrid","Ygrid","IDNgrid")],
FUN = length)
Totalgrids <- length(levels(bcidata$IDgrid))
Occupiedgrids <- length(aggregate$IDNgrid)
sum <- sum(aggregate$x)
TotalArea <- 50
Area <- (1000/i*500/j)
Occupancy <- (Occupiedgrids/Totalgrids)
Mean <- length(bcidata$Latin)/(Occupiedgrids)
Variance <- var(aggregate$x)
occupancymean.data.frame <- rbind(occupancymean.data.frame,
data.frame(Area, Totalgrids, Occupiedgrids, Occupancy, Mean, Variance))
}
occupancymean.data.frame
Occupancy <- occupancymean.data.frame$Occupancy
Mean <- occupancymean.data.frame$Mean
poission <- nls(Occupancy ~ 1-exp(-rho*Mean), start = list(rho = 2.1),
data = occupancymean.data.frame)
nachman <- nls(Occupancy ~ 1-exp(-alpha*Mean^beta), start = list(alpha =
0.2, beta = 0.1), data = occupancymean.data.frame)
logistic <- nls(Occupancy ~ (alpha*Mean^beta)/(1+alpha*Mean^beta), start
= list(alpha = 0.2, beta = 0.1),data = occupancymean.data.frame)
nbd <- nls(Occupancy ~ 1-(1+(Mean)/k)^-k, start = list(k = 1), data =
occupancymean.data.frame)
power <- nls(Occupancy ~ alpha*Mean^beta, start = list(alpha = 0.2, beta=
0.1), data = occupancymean.data.frame)
inbd <- nls(Occupancy ~
1-(alpha*(Mean^(beta-1)))^(Mean/(1-alpha*Mean^(beta-1))), start =
list(alpha = 0.2, beta = 0.3),
data = occupancymean.data.frame)
fnbd <- nls(Occupancy ~ 1- (gamma(N +
k/(Mean*A/N)-k)*gamma(k/(Mean*A/N)))/(gamma(k/(Mean*A/N)-k)*gamma(N+k/(Mean*A/N))),
start = list(k = 0.2, A = 0.1, N = 0.2), data = occupancymean
.data.frame)
bayesianII <- nls(Occupancy ~ 1-(theta*beta^(2*(TotalArea
*Mean/sum)^0.5)*delta^(TotalArea*Mean/sum)), start = list(theta=0.9956,
beta=1, delta=1), data = occupancymean.data.frame)
return(list(summary(poission), summary(nachman), summary(logistic),
summary(nbd),
summary(power), summary(inbd), summary(fnbd), summary(
bayesianII)))
}
Modelsfunc(bcidata)
Your assistance will be highly appreciated. Thanks in advance.
Regards,
*Evans Ochiaga*
*African Institute for Mathematical Sciences*
*6 Melrose Road*
*Muizenberg, South Africa*
*Msc in Mathematical Sciences+27 84 61 69 183 *
*"When I cannot understand my Father’s leading, And it seems to be but hard
and cruel fate, Still I hear that gentle whisper ever pleading, God is
working, God is faithful—Only wait."*
[[alternative HTML version deleted]]
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] Asssistance
On 03/03/2015 5:47 AM, Evans Otieno Ochiaga wrote:
> Hi to All,
>
> I am building a package in R and whenever I run command "R CMD build OAR"
> in the terminal, I get the following error:
>
> * checking for file ‘OAR/DESCRIPTION’ ... OK
> * preparing ‘OAR’:
> * checking DESCRIPTION meta-information ... ERROR
> Malformed Depends or Suggests or Imports or Enhances field.
> Offending entries:
> R (>=3.0.2)
> Entries must be names of packages optionally followed by '<=' or '>=',
> white space, and a valid version number in parentheses.
That looks okay; I'm guessing it is out of place. Can you show use your
DESCRIPTION file?
Duncan Murdoch
>
> See the information on DESCRIPTION files in section 'Creating R
> packages' of the 'Writing R Extensions' manual.
>
> This is my first time to build a package using R and it's very hard for me
> to figure out where the problem is. I kindly call for your assistance in
> fixing the problem. Below is my function;
>
> bcidata <- read.csv("~/Desktop/Files_for_Package/data.csv"); bcidata
>
> Modelsfunc<- function(bcidata){
>
> occupancymean.data.frame <- NULL
>
> for (k in seq(2.5,250,by=2.5)){
>
> i <- 1000/k
>
> j <- 500/k
>
> bcidata$Xgrid <- cut(bcidata$PX, breaks = i, include.lowest = T)
>
> bcidata$Ygrid <- cut(bcidata$PY, breaks = j, include.lowest = T)
>
> bcidata$IDgrid <- with(bcidata, interaction(Xgrid,Ygrid))
>
> bcidata$IDNgrid <- factor(bcidata$IDgrid)
>
> levels(bcidata$IDgrid) <- seq_along(levels(bcidata$IDgrid))
>
> bcidata$count <- ave(bcidata$PX, bcidata$IDgrid, FUN = length)
>
> aggregate <- aggregate(bcidata$PX,bcidata[,c("Xgrid","Ygrid","IDNgrid")],
> FUN = length)
>
> Totalgrids <- length(levels(bcidata$IDgrid))
>
> Occupiedgrids <- length(aggregate$IDNgrid)
>
> sum <- sum(aggregate$x)
>
> TotalArea <- 50
>
> Area <- (1000/i*500/j)
>
> Occupancy <- (Occupiedgrids/Totalgrids)
>
> Mean <- length(bcidata$Latin)/(Occupiedgrids)
>
> Variance <- var(aggregate$x)
>
> occupancymean.data.frame <- rbind(occupancymean.data.frame,
> data.frame(Area, Totalgrids, Occupiedgrids, Occupancy, Mean, Variance))
>
> }
>
> occupancymean.data.frame
>
> Occupancy <- occupancymean.data.frame$Occupancy
>
> Mean <- occupancymean.data.frame$Mean
>
> poission <- nls(Occupancy ~ 1-exp(-rho*Mean), start = list(rho = 2.1),
> data = occupancymean.data.frame)
>
> nachman <- nls(Occupancy ~ 1-exp(-alpha*Mean^beta), start = list(alpha =
> 0.2, beta = 0.1), data = occupancymean.data.frame)
>
> logistic <- nls(Occupancy ~ (alpha*Mean^beta)/(1+alpha*Mean^beta), start
> = list(alpha = 0.2, beta = 0.1),data = occupancymean.data.frame)
>
> nbd <- nls(Occupancy ~ 1-(1+(Mean)/k)^-k, start = list(k = 1), data =
> occupancymean.data.frame)
>
> power <- nls(Occupancy ~ alpha*Mean^beta, start = list(alpha = 0.2, beta=
> 0.1), data = occupancymean.data.frame)
>
> inbd <- nls(Occupancy ~
> 1-(alpha*(Mean^(beta-1)))^(Mean/(1-alpha*Mean^(beta-1))), start =
> list(alpha = 0.2, beta = 0.3),
>
> data = occupancymean.data.frame)
>
> fnbd <- nls(Occupancy ~ 1- (gamma(N +
> k/(Mean*A/N)-k)*gamma(k/(Mean*A/N)))/(gamma(k/(Mean*A/N)-k)*gamma(N+k/(Mean*A/N))),
>
>
> start = list(k = 0.2, A = 0.1, N = 0.2), data = occupancymean
> .data.frame)
>
> bayesianII <- nls(Occupancy ~ 1-(theta*beta^(2*(TotalArea
> *Mean/sum)^0.5)*delta^(TotalArea*Mean/sum)), start = list(theta=0.9956,
> beta=1, delta=1), data = occupancymean.data.frame)
>
>
> return(list(summary(poission), summary(nachman), summary(logistic),
> summary(nbd),
>
> summary(power), summary(inbd), summary(fnbd), summary(
> bayesianII)))
>
> }
>
> Modelsfunc(bcidata)
>
> Your assistance will be highly appreciated. Thanks in advance.
>
> Regards,
>
>
> *Evans Ochiaga*
>
> *African Institute for Mathematical Sciences*
>
> *6 Melrose Road*
>
> *Muizenberg, South Africa*
>
> *Msc in Mathematical Sciences+27 84 61 69 183 *
>
> *"When I cannot understand my Father’s leading, And it seems to be but hard
> and cruel fate, Still I hear that gentle whisper ever pleading, God is
> working, God is faithful—Only wait."*
>
> [[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
Re: [Rd] Asssistance
Hi Evans, > * checking for file ‘OAR/DESCRIPTION’ ... OK > * preparing ‘OAR’: > * checking DESCRIPTION meta-information ... ERROR > Malformed Depends or Suggests or Imports or Enhances field. > Offending entries: > R (>=3.0.2) > Entries must be names of packages optionally followed by '<=' or '>=', > white space, and a valid version number in parentheses. The _white space_ (see explanation above) seems to be missing. Try "R (>= 3.0.2)" Best, /g __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] Import data set from another package?
As I expected: there was something simple and obvious, which I somehow could
not see.
Thanks for the pointer.
Terry T.
On 03/03/2015 03:12 AM, Prof Brian Ripley wrote:
On 02/03/2015 22:48, Therneau, Terry M., Ph.D. wrote:
I've moved nlme from Depends to Imports in my coxme package. However, a
few of the examples for lmekin use one of the data sets from nlme. This
is on purpose, to show how the results are the same and how they differ.
If I use data(nlme::ergoStool) the data is not found,
data(nlme:::ergoStool) does no better.
If I add importFrom(nlme, "ergoStool") the error message is that
ergoStool is not exported.
There likely is a simple way, but I currently don't see it.
There were some off-the-mark suggestions in this thread. If you just want a
dataset from
a package, use
data("ergoStool", package = "nlme")
In particular, it is somewhat wasteful to load a large namespace like nlme when
it is not
needed.
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] Asking for tasks of summer code 2015
On Mo, 2015-03-02 at 16:53 +0100, han cao wrote: > Hey everyone: > I am a Master student from Saarland Unirversity, Germany with the major of > Bioinformatics. And I am interested in statistical learning which is also > my major work in the future with the implementation by R. So I 'd like join > the google summer code this year by doing tasks in your community. However > I can not find whether there are tasks available provided for this year, > anyone can tell me? The R GSoC pages moved to GitHub: https://github.com/rstats-gsoc/gsoc2015/wiki Best, Claudia > > > Hank Cao > > [[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
[Rd] Feature request: copy attributes in gzcon
The `gzcon` function both modifies and copies a connection object:
# compressed text
con1 <- url("http://www.stats.ox.ac.uk/pub/datasets/csb/ch12.dat.gz";)
con2 <- gzcon(con1)
# almost indistinguishable
con1==con2
identical(summary(con2), summary(con1))
# both support gzip
readLines(con1, n = 3)
readLines(con2, n = 3)
# opening one opens both
isOpen(con2)
open(con1)
isOpen(con2)
In the example, `con1` and `con2` are two different objects
interfacing the same connection. It might seem as if gzcon has simply
returned the modified connection object, but the documentation
explains that it in fact creates a copy referencing the same
connection but with a "modified internal structure".
It is unclear to me how `con1` is different from `con2`, but given
that they represent one and the same connection, would it be possible
to make gzcon copy over attributes from the input connection to the
output object?
This would allow custom connection implementations such as the curl
package to use attributes for storing additional metadata about
connection. Currently those attributes get dropped after calling gzcon
on the connection:
library(curl)
con <- curl("http://www.stats.ox.ac.uk/pub/datasets/csb/ch12.dat.gz";)
attr(con, "foo") <- "bar"
con <- gzcon(con)
attr(con, "foo")
It would be very helpful if gzcon would instead copy attributes onto
the output object, such that any potential meta-data about the
connection as stored in attributes gets retained.
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] [R] Why does R replace all row values with NAs
On 03/03/2015 02:28 AM, Martin Maechler wrote:
Diverted from R-help :
as it gets into musing about new R language "primitives"
William Dunlap
on Fri, 27 Feb 2015 08:04:36 -0800 writes:
> You could define functions like
> is.true <- function(x) !is.na(x) & x
> is.false <- function(x) !is.na(x) & !x
> and use them in your selections. E.g.,
>> x <- data.frame(a=1:10,b=2:11,c=c(1,NA,3,NA,5,NA,7,NA,NA,10))
>> x[is.true(x$c >= 6), ]
> a b c
> 7 7 8 7
> 10 10 11 10
> Bill Dunlap
> TIBCO Software
> wdunlap tibco.com
Yes; the Matrix package has had these
is0 <- function(x) !is.na(x) & x == 0
isN0 <- function(x) is.na(x) | x != 0
is1 <- function(x) !is.na(x) & x # also == "isTRUE componentwise"
Note that using %in% to block propagation of NAs is about 2x faster:
> x <- sample(c(NA_integer_, 1:1), 50, replace=TRUE)
> microbenchmark(as.logical(x) %in% TRUE, !is.na(x) & x)
Unit: milliseconds
expr minlq mean medianuq
as.logical(x) %in% TRUE 6.034744 6.264382 6.999083 6.29488 6.346028
!is.na(x) & x 11.202808 11.402437 11.469101 11.44848 11.517576
max neval
40.36472 100
11.90916 100
namespace hidden for a while [note the comment of the last one!]
and using them for readibility in its own code.
Maybe we should (again) consider providing some versions of
these with R ?
The Matrix package also has had fast
allFalse <- all0 <- function(x) .Call(R_all0, x)
anyFalse <- any0 <- function(x) .Call(R_any0, x)
##
## anyFalse <- function(x) isTRUE(any(!x))## ~= any0
## any0 <- function(x) isTRUE(any(x == 0)) ## ~= anyFalse
namespace hidden as well, already, which probably could also be
brought to base R.
One big reason to *not* go there (to internal C code) at all with R is that
S3 and S4 dispatch for '==' ('!=', etc, the 'Compare' group generics)
and 'is.na() have been known and package writers have
programmed methods for these.
To ensure that S3 and S4 dispatch works "correctly" also inside
such new internals is much less easily achieved, and so
such a C-based internal function is0() would no longer be
equivalent with!is.na(x) & x == 0
as soon as 'x' is an "object" with a '==', 'Compare' and/or an is.na() method.
Excellent point. Thank you! It really makes a big difference for
developers who maintain a complex hierarchy of S4 classes and methods,
when functions like is.true, anyFalse, etc..., which can be expressed in
terms of more basic operations like ==, !=, !, is.na, etc..., just work
out-of-the-box on objects for which these basic operations are defined.
There is conceptually a small set of "building blocks", at least for
objects with a vector-like or list-like semantic, that can be used
to formally describe the semantic of many functions in base R. This
is what the man page for anyNA does by saying:
anyNA implements any(is.na(x))
even though the actual implementation differs, but that's ok, as long
as anyNA is equivalent to doing any(is.na(x)) on any object for which
building block is.na() is implemented.
Unfortunately there is no clearly identified set of building blocks
in base R. For example, if I want the comparison operations to work
on my object, I need to implement ==, >, <, !=, <=, and >= (the
'Compare' group generics) even though it should be enough to implement
== and >=, because all the others can be described in terms of these
2 building blocks. unique/duplicated is another example (unique(x) is
conceptually x[!duplicated(x)]). And so on...
Cheers,
H.
OTOH, simple R versions such as your 'is.true', called 'is1'
inside Matrix maybe optimizable a bit by the byte compiler (and
jit and other such tricks) and still keep the full
semantic including correct method dispatch.
Martin Maechler, ETH Zurich
> On Fri, Feb 27, 2015 at 7:27 AM, Dimitri Liakhovitski <
> dimitri.liakhovit...@gmail.com> wrote:
>> Thank you very much, Duncan.
>> All this being said:
>>
>> What would you say is the most elegant and most safe way to solve such
>> a seemingly simple task?
>>
>> Thank you!
>>
>> On Fri, Feb 27, 2015 at 10:02 AM, Duncan Murdoch
>> wrote:
>> > On 27/02/2015 9:49 AM, Dimitri Liakhovitski wrote:
>> >> So, Duncan, do I understand you correctly:
>> >>
>> >> When I use x$x<6, R doesn't know if it's TRUE or FALSE, so it returns
>> >> a logical value of NA.
>> >
>> > Yes, when x$x is NA. (Though I think you meant x$c.)
>> >
>> >> When this logical value is applied to a row, the R says: hell, I
don't
>> >> know if I should keep it or not, so, just in case, I am going to keep
>> >> it, but I'll replace all the values in this row with NAs?
>> >
>> > Yes. Indexing with a logical NA is probably a mistake, and this is
one
>> > way to signal it without actually trigge
Re: [Rd] [R] Why does R replace all row values with NAs
On 3/3/15 1:26 PM, Hervé Pagès wrote:
On 03/03/2015 02:28 AM, Martin Maechler wrote:
Diverted from R-help :
as it gets into musing about new R language "primitives"
William Dunlap
on Fri, 27 Feb 2015 08:04:36 -0800 writes:
> You could define functions like
> is.true <- function(x) !is.na(x) & x
> is.false <- function(x) !is.na(x) & !x
> and use them in your selections. E.g.,
>> x <- data.frame(a=1:10,b=2:11,c=c(1,NA,3,NA,5,NA,7,NA,NA,10))
>> x[is.true(x$c >= 6), ]
> a b c
> 7 7 8 7
> 10 10 11 10
> Bill Dunlap
> TIBCO Software
> wdunlap tibco.com
Yes; the Matrix package has had these
is0 <- function(x) !is.na(x) & x == 0
isN0 <- function(x) is.na(x) | x != 0
is1 <- function(x) !is.na(x) & x # also == "isTRUE componentwise"
Note that using %in% to block propagation of NAs is about 2x faster:
> x <- sample(c(NA_integer_, 1:1), 50, replace=TRUE)
> microbenchmark(as.logical(x) %in% TRUE, !is.na(x) & x)
Unit: milliseconds
expr minlq mean medianuq
as.logical(x) %in% TRUE 6.034744 6.264382 6.999083 6.29488 6.346028
!is.na(x) & x 11.202808 11.402437 11.469101 11.44848 11.517576
max neval
40.36472 100
11.90916 100
Unfortunately %in% does not preserve matrix dimensions:
> x <- matrix(sample(c(NA_integer_, 1:100), 500, replace=TRUE), nrow=50)
> dim(x)
[1] 50 10
> dim(!is.na(x) & x)
[1] 50 10
> dim(as.logical(x) %in% TRUE)
NULL
Stephanie
namespace hidden for a while [note the comment of the last one!]
and using them for readibility in its own code.
Maybe we should (again) consider providing some versions of
these with R ?
The Matrix package also has had fast
allFalse <- all0 <- function(x) .Call(R_all0, x)
anyFalse <- any0 <- function(x) .Call(R_any0, x)
##
## anyFalse <- function(x) isTRUE(any(!x)) ## ~= any0
## any0 <- function(x) isTRUE(any(x == 0)) ## ~= anyFalse
namespace hidden as well, already, which probably could also be
brought to base R.
One big reason to *not* go there (to internal C code) at all with R is
that
S3 and S4 dispatch for '==' ('!=', etc, the 'Compare' group generics)
and 'is.na() have been known and package writers have
programmed methods for these.
To ensure that S3 and S4 dispatch works "correctly" also inside
such new internals is much less easily achieved, and so
such a C-based internal function is0() would no longer be
equivalent with!is.na(x) & x == 0
as soon as 'x' is an "object" with a '==', 'Compare' and/or an is.na()
method.
Excellent point. Thank you! It really makes a big difference for
developers who maintain a complex hierarchy of S4 classes and methods,
when functions like is.true, anyFalse, etc..., which can be expressed in
terms of more basic operations like ==, !=, !, is.na, etc..., just work
out-of-the-box on objects for which these basic operations are defined.
There is conceptually a small set of "building blocks", at least for
objects with a vector-like or list-like semantic, that can be used
to formally describe the semantic of many functions in base R. This
is what the man page for anyNA does by saying:
anyNA implements any(is.na(x))
even though the actual implementation differs, but that's ok, as long
as anyNA is equivalent to doing any(is.na(x)) on any object for which
building block is.na() is implemented.
Unfortunately there is no clearly identified set of building blocks
in base R. For example, if I want the comparison operations to work
on my object, I need to implement ==, >, <, !=, <=, and >= (the
'Compare' group generics) even though it should be enough to implement
== and >=, because all the others can be described in terms of these
2 building blocks. unique/duplicated is another example (unique(x) is
conceptually x[!duplicated(x)]). And so on...
Cheers,
H.
OTOH, simple R versions such as your 'is.true', called 'is1'
inside Matrix maybe optimizable a bit by the byte compiler (and
jit and other such tricks) and still keep the full
semantic including correct method dispatch.
Martin Maechler, ETH Zurich
> On Fri, Feb 27, 2015 at 7:27 AM, Dimitri Liakhovitski <
> dimitri.liakhovit...@gmail.com> wrote:
>> Thank you very much, Duncan.
>> All this being said:
>>
>> What would you say is the most elegant and most safe way to
solve such
>> a seemingly simple task?
>>
>> Thank you!
>>
>> On Fri, Feb 27, 2015 at 10:02 AM, Duncan Murdoch
>> wrote:
>> > On 27/02/2015 9:49 AM, Dimitri Liakhovitski wrote:
>> >> So, Duncan, do I understand you correctly:
>> >>
>> >> When I use x$x<6, R doesn't know if it's TRUE or FALSE, so
it returns
>> >> a logical value of NA.
>> >
>> > Yes, when x$x is NA. (Though I think you meant x$c.)
>> >
>> >> When this logical value is applied to a row, the R says:
hell, I don't
>> >> know i
Re: [Rd] [R] Why does R replace all row values with NAs
Stephanie,
Actually, it's as.logical that isn't preserving matrix dimensions, because
it coerces to a logical vector:
> x <- matrix(sample(c(NA_integer_, 1:100), 500, replace=TRUE), nrow=50)
> dim(as.logical(x))
NULL
~G
On Tue, Mar 3, 2015 at 2:09 PM, Stephanie M. Gogarten <
sdmor...@u.washington.edu> wrote:
>
>
> On 3/3/15 1:26 PM, Hervé Pagès wrote:
>
>>
>>
>> On 03/03/2015 02:28 AM, Martin Maechler wrote:
>>
>>> Diverted from R-help :
>>> as it gets into musing about new R language "primitives"
>>>
>>> William Dunlap
on Fri, 27 Feb 2015 08:04:36 -0800 writes:
>>>
>>> > You could define functions like
>>>
>>> > is.true <- function(x) !is.na(x) & x
>>> > is.false <- function(x) !is.na(x) & !x
>>>
>>> > and use them in your selections. E.g.,
>>> >> x <- data.frame(a=1:10,b=2:11,c=c(1,NA,3,NA,5,NA,7,NA,NA,10))
>>> >> x[is.true(x$c >= 6), ]
>>> > a b c
>>> > 7 7 8 7
>>> > 10 10 11 10
>>>
>>> > Bill Dunlap
>>> > TIBCO Software
>>> > wdunlap tibco.com
>>>
>>> Yes; the Matrix package has had these
>>>
>>> is0 <- function(x) !is.na(x) & x == 0
>>> isN0 <- function(x) is.na(x) | x != 0
>>> is1 <- function(x) !is.na(x) & x # also == "isTRUE componentwise"
>>>
>>
>> Note that using %in% to block propagation of NAs is about 2x faster:
>>
>> > x <- sample(c(NA_integer_, 1:1), 50, replace=TRUE)
>> > microbenchmark(as.logical(x) %in% TRUE, !is.na(x) & x)
>> Unit: milliseconds
>> expr minlq mean medianuq
>> as.logical(x) %in% TRUE 6.034744 6.264382 6.999083 6.29488 6.346028
>> !is.na(x) & x 11.202808 11.402437 11.469101 11.44848
>> 11.517576
>>max neval
>> 40.36472 100
>> 11.90916 100
>>
>
> Unfortunately %in% does not preserve matrix dimensions:
>
> > x <- matrix(sample(c(NA_integer_, 1:100), 500, replace=TRUE), nrow=50)
> > dim(x)
> [1] 50 10
> > dim(!is.na(x) & x)
> [1] 50 10
> > dim(as.logical(x) %in% TRUE)
> NULL
>
> Stephanie
>
>
>
>>
>>
>>
>>> namespace hidden for a while [note the comment of the last one!]
>>> and using them for readibility in its own code.
>>>
>>> Maybe we should (again) consider providing some versions of
>>> these with R ?
>>>
>>> The Matrix package also has had fast
>>>
>>> allFalse <- all0 <- function(x) .Call(R_all0, x)
>>> anyFalse <- any0 <- function(x) .Call(R_any0, x)
>>> ##
>>> ## anyFalse <- function(x) isTRUE(any(!x)) ## ~= any0
>>> ## any0 <- function(x) isTRUE(any(x == 0)) ## ~= anyFalse
>>>
>>> namespace hidden as well, already, which probably could also be
>>> brought to base R.
>>>
>>> One big reason to *not* go there (to internal C code) at all with R is
>>> that
>>> S3 and S4 dispatch for '==' ('!=', etc, the 'Compare' group generics)
>>> and 'is.na() have been known and package writers have
>>> programmed methods for these.
>>> To ensure that S3 and S4 dispatch works "correctly" also inside
>>> such new internals is much less easily achieved, and so
>>> such a C-based internal function is0() would no longer be
>>> equivalent with!is.na(x) & x == 0
>>> as soon as 'x' is an "object" with a '==', 'Compare' and/or an is.na()
>>> method.
>>>
>>
>> Excellent point. Thank you! It really makes a big difference for
>> developers who maintain a complex hierarchy of S4 classes and methods,
>> when functions like is.true, anyFalse, etc..., which can be expressed in
>> terms of more basic operations like ==, !=, !, is.na, etc..., just work
>> out-of-the-box on objects for which these basic operations are defined.
>>
>> There is conceptually a small set of "building blocks", at least for
>> objects with a vector-like or list-like semantic, that can be used
>> to formally describe the semantic of many functions in base R. This
>> is what the man page for anyNA does by saying:
>>
>>anyNA implements any(is.na(x))
>>
>> even though the actual implementation differs, but that's ok, as long
>> as anyNA is equivalent to doing any(is.na(x)) on any object for which
>> building block is.na() is implemented.
>>
>> Unfortunately there is no clearly identified set of building blocks
>> in base R. For example, if I want the comparison operations to work
>> on my object, I need to implement ==, >, <, !=, <=, and >= (the
>> 'Compare' group generics) even though it should be enough to implement
>> == and >=, because all the others can be described in terms of these
>> 2 building blocks. unique/duplicated is another example (unique(x) is
>> conceptually x[!duplicated(x)]). And so on...
>>
>> Cheers,
>> H.
>>
>>
>>> OTOH, simple R versions such as your 'is.true', called 'is1'
>>> inside Matrix maybe optimizable a bit by the byte compiler (and
>>> jit and other such tricks) and still keep the full
>>> semantic including correct method dispatch.
>>>
>>> Martin Maechler, ETH Zurich
>>>
>>>
>>> > On Fri, Feb 27, 2015 at 7:27 AM, Dimitri Liakhovitski <
>>> > dimitri.liakhov
Re: [Rd] [R] Why does R replace all row values with NAs
On 03/03/2015 02:17 PM, Gabriel Becker wrote:
Stephanie,
Actually, it's as.logical that isn't preserving matrix dimensions,
because it coerces to a logical vector:
> x <- matrix(sample(c(NA_integer_, 1:100), 500, replace=TRUE), nrow=50)
> dim(as.logical(x))
It's true, as.logical() doesn't help here but Stephanie is right, %in%
does not preserve the dimensions either:
> dim(x %in% 1:5)
NULL
That's because match() itself doesn't preserve the dimensions:
> dim(match(x, 1:5))
NULL
So maybe my fast is.true() should be:
is.true <- function(x)
{
ans <- as.logical(x) %in% TRUE
if (is.null(dim(x))) {
names(ans) <- names(x)
} else {
dim(ans) <- dim(x)
dimnames(ans) <- dimnames(x)
}
ans
}
or something like that...
H.
NULL
~G
On Tue, Mar 3, 2015 at 2:09 PM, Stephanie M. Gogarten
mailto:sdmor...@u.washington.edu>> wrote:
On 3/3/15 1:26 PM, Hervé Pagès wrote:
On 03/03/2015 02:28 AM, Martin Maechler wrote:
Diverted from R-help :
as it gets into musing about new R language "primitives"
William Dunlap mailto:wdun...@tibco.com>>
on Fri, 27 Feb 2015 08:04:36 -0800
writes:
> You could define functions like
> is.true <- function(x) !is.na (x) & x
> is.false <- function(x) !is.na (x) & !x
> and use them in your selections. E.g.,
>> x <-
data.frame(a=1:10,b=2:11,c=c(__1,NA,3,NA,5,NA,7,NA,NA,10))
>> x[is.true(x$c >= 6), ]
> a b c
> 7 7 8 7
> 10 10 11 10
> Bill Dunlap
> TIBCO Software
> wdunlap tibco.com
Yes; the Matrix package has had these
is0 <- function(x) !is.na (x) & x == 0
isN0 <- function(x) is.na (x) | x != 0
is1 <- function(x) !is.na (x) & x # also ==
"isTRUE componentwise"
Note that using %in% to block propagation of NAs is about 2x faster:
> x <- sample(c(NA_integer_, 1:1), 50, replace=TRUE)
> microbenchmark(as.logical(x) %in% TRUE, !is.na
(x) & x)
Unit: milliseconds
expr minlq mean
medianuq
as.logical(x) %in% TRUE 6.034744 6.264382 6.999083
6.29488 6.346028
!is.na (x) & x 11.202808 11.402437
11.469101 11.44848 11.517576
max neval
40.36472 100
11.90916 100
Unfortunately %in% does not preserve matrix dimensions:
> x <- matrix(sample(c(NA_integer_, 1:100), 500, replace=TRUE),
nrow=50)
> dim(x)
[1] 50 10
> dim(!is.na (x) & x)
[1] 50 10
> dim(as.logical(x) %in% TRUE)
NULL
Stephanie
namespace hidden for a while [note the comment of the last
one!]
and using them for readibility in its own code.
Maybe we should (again) consider providing some versions of
these with R ?
The Matrix package also has had fast
allFalse <- all0 <- function(x) .Call(R_all0, x)
anyFalse <- any0 <- function(x) .Call(R_any0, x)
##
## anyFalse <- function(x) isTRUE(any(!x)) ## ~= any0
## any0 <- function(x) isTRUE(any(x == 0)) ## ~=
anyFalse
namespace hidden as well, already, which probably could also be
brought to base R.
One big reason to *not* go there (to internal C code) at all
with R is
that
S3 and S4 dispatch for '==' ('!=', etc, the 'Compare' group
generics)
and 'is.na () have been known and package
writers have
programmed methods for these.
To ensure that S3 and S4 dispatch works "correctly" also inside
such new internals is much less easily achieved, and so
such a C-based internal function is0() would no longer be
equivalent with!is.na (x) & x == 0
as soon as 'x' is an "object" with a '==', 'Compare' and/or
an is.na ()
method.
Excellent point. Thank you! It really makes a big difference for
developers who maintain a complex hierarchy of S4 classes and
methods,
when functions like is.true, anyFalse, etc..., which can be
expressed in
terms of more basic operations like ==, !=, !, is.na
, etc..., just work
out-of-the-box on objects for which these basic operations are
de
Re: [Rd] Errors on Windows with grep(fixed=TRUE) on UTF-8 strings
After a bit more investigation, I think I've found the cause of the bug,
and I have a patch.
This bug happens with grep(), when:
* Running on Windows.
* The search uses fixed=TRUE.
* The search pattern is a single byte.
* The current locale has a multibyte encoding.
===
Here's an example that demonstrates the bug:
# First, create a 3-byte UTF-8 character
y <- rawToChar(as.raw(c(0xe6, 0xb8, 0x97)))
Encoding(y) <- "UTF-8"
y
# [1] "渗"
# In my default locale, grep with a single-char pattern and fixed=TRUE
# returns integer(0), as expected.
Sys.getlocale("LC_CTYPE")
# [1] "English_United States.1252"
grep("a", y, fixed = TRUE)
# integer(0)
# When the using a multibyte locale, grep with a single-char
# pattern and fixed=TRUE results in an error.
Sys.setlocale("LC_CTYPE", "chinese")
grep("a", y, fixed = TRUE)
# Error in grep("a", y, fixed = TRUE) : invalid multibyte string at '<97>'
===
I believe the problem is in the main/grep.c file, in the fgrep_one
function. It tests for a multi-byte character string locale
`mbcslocale`, and then for the `use_UTF8`, like so:
if (!useBytes && mbcslocale) {
...
} else if (!useBytes && use_UTF8) {
...
} else ...
This can be seen at
https://github.com/wch/r-source/blob/e92b4c1cba05762480cd3898335144e5dd111cb7/src/main/grep.c#L668-L692
A similar pattern occurs in the fgrep_one_bytes function, at
https://github.com/wch/r-source/blob/e92b4c1cba05762480cd3898335144e5dd111cb7/src/main/grep.c#L718-L736
I believe that the test order should be reversed; it should test first
for `use_UTF8`, and then for `mbcslocale`. This pattern occurs in a few
places in grep.c. It looks like this:
if (!useBytes && use_UTF8) {
...
} else if (!useBytes && mbcslocale) {
...
} else ...
===
This patch does what I described; it simply tests for `use_UTF8` first,
and then `mbcslocale`, in both fgrep_one and fgrep_one_bytes. I made
this patch against the 3.1.2 sources, and tested the example code above.
In both cases, grep() returned integer(0), as expected.
(The reason I made this change against 3.1.2 is because I had problems
getting the current trunk to compile on both Linux or Windows.)
diff --git src/main/grep.c src/main/grep.c
index 6e6ec3e..348c63d 100644
--- src/main/grep.c
+++ src/main/grep.c
@@ -664,27 +664,27 @@ static int fgrep_one(const char *pat, const char *target,
}
return -1;
}
-if (!useBytes && mbcslocale) { /* skip along by chars */
- mbstate_t mb_st;
+if (!useBytes && use_UTF8) {
int ib, used;
- mbs_init(&mb_st);
for (ib = 0, i = 0; ib <= len-plen; i++) {
if (strncmp(pat, target+ib, plen) == 0) {
if (next != NULL) *next = ib + plen;
return i;
}
- used = (int) Mbrtowc(NULL, target+ib, MB_CUR_MAX, &mb_st);
+ used = utf8clen(target[ib]);
if (used <= 0) break;
ib += used;
}
-} else if (!useBytes && use_UTF8) {
+} else if (!useBytes && mbcslocale) { /* skip along by chars */
+ mbstate_t mb_st;
int ib, used;
+ mbs_init(&mb_st);
for (ib = 0, i = 0; ib <= len-plen; i++) {
if (strncmp(pat, target+ib, plen) == 0) {
if (next != NULL) *next = ib + plen;
return i;
}
- used = utf8clen(target[ib]);
+ used = (int) Mbrtowc(NULL, target+ib, MB_CUR_MAX, &mb_st);
if (used <= 0) break;
ib += used;
}
@@ -714,21 +714,21 @@ static int fgrep_one_bytes(const char *pat, const char
*target, int len,
if (*p == pat[0]) return i;
return -1;
}
-if (!useBytes && mbcslocale) { /* skip along by chars */
- mbstate_t mb_st;
+if (!useBytes && use_UTF8) { /* not really needed */
int ib, used;
- mbs_init(&mb_st);
for (ib = 0, i = 0; ib <= len-plen; i++) {
if (strncmp(pat, target+ib, plen) == 0) return ib;
- used = (int) Mbrtowc(NULL, target+ib, MB_CUR_MAX, &mb_st);
+ used = utf8clen(target[ib]);
if (used <= 0) break;
ib += used;
}
-} else if (!useBytes && use_UTF8) { /* not really needed */
+} else if (!useBytes && mbcslocale) { /* skip along by chars */
+ mbstate_t mb_st;
int ib, used;
+ mbs_init(&mb_st);
for (ib = 0, i = 0; ib <= len-plen; i++) {
if (strncmp(pat, target+ib, plen) == 0) return ib;
- used = utf8clen(target[ib]);
+ used = (int) Mbrtowc(NULL, target+ib, MB_CUR_MAX, &mb_st);
if (used <= 0) break;
ib += used;
}
-Winston
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
