Re: [Rd] Roxygen
>> It is desirable that these tags and their syntax and semantics be >> compatible with those used by Doxygen and Javadoc, but extended with >> tags important for R specific things. > > Doxygen, in particular, has some tags that are meaningless in R > (@category and @protocol come to mind); wouldn't it be worthwhile, > therefore, to start with a subset of Rd? > > Thinks like @name and @format could be gleaned from the code; and it > seems like preserving other Rd names would uphold the "principle of > least surprise." > Yes, actually I did not meant that you have to take the union of Javadoc and Doxygen keywords and try to fit it into R. Start with the things from R documentation, extend it with usefull R-specific keywords (like @export). And keep in mind that there may be users which use Doxygen and Javadoc and we want to uphold the "principle of least surprise". :-) Manuel. __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] Problems building rpm for R-2.6.2 on Opensuse Factory
On Tue, 18 Mar 2008 10:30:07 + (GMT) Prof Brian Ripley <[EMAIL PROTECTED]> wrote: > This is proving elusive. The same build that gave incorrect results last > week gave correct results last night (and F8 has had about 50 updates > since, including two kernel updates). > > I'd still suggest trying lower levels of optimization, as that was > definitely needed with some pre-releases of gcc 4.3.x. For the record: CFLAGS="-O" did it this time. detlef > > > On Mon, 17 Mar 2008, Prof Brian Ripley wrote: > > > On Mon, 17 Mar 2008, Detlef Steuer wrote: > > > >> Hi, > >> > >> make check fails in the way shown below when I try to build > >> R-2.6.2 on OpenSuSE Factory x86_64. It builds fine on i586. > >> (and on all older Versions of Opensuse) > >> > >> Is there something I can do? > >> Are there known problems with some gcc versions? > >> (Factory uses 4.3.0) > > > > Yes. This happened for me with (vanilla released) gcc 4.3.0 on F8 x86_64 > > too. Turning optimization off for both C and Fortran solved it -- I've not > > yet had time to track down what you need not to optimize. I'll try to do > > that this evening. > > > > As I understand it, CRAN is successfully building R with a Debian x86_64 > > build of gcc 4.3.0, but several packages are failing in that configuration. > > > >> > >> Output from ./configure is > >> - > >> R is now configured for x86_64-unknown-linux-gnu > >> > >> Source directory: . > >> Installation directory: /usr > >> > >> C compiler: gcc -std=gnu99 -g -O2 > >> Fortran 77 compiler: gfortran -g -O2 > >> > >> C++ compiler: g++ -g -O2 > >> Fortran 90/95 compiler: gfortran -g -O2 > >> Obj-C compiler: > >> > >> Interfaces supported: X11, tcltk > >> External libraries: readline > >> Additional capabilities: PNG, JPEG, iconv, MBCS, NLS > >> Options enabled: shared R library, shared BLAS, R profiling > >> > >> Recommended packages: yes > >> - > >> > >> It may very well be an error on the Opensuse side, but I'm just not > >> sure how to identify the true cause of the problem. > >> Any help appreciated. > >> > >> Detlef > >> > >> - > >> Error message: > >> > >> running code in 'lapack.R' ... OK > >> comparing 'lapack.Rout' to './lapack.Rout.save' ... OK > >> make[3]: Leaving directory `/usr/src/packages/BUILD/R-2.6.2/tests' > >> make[2]: Leaving directory `/usr/src/packages/BUILD/R-2.6.2/tests' > >> make[2]: Entering directory `/usr/src/packages/BUILD/R-2.6.2/tests' > >> running regression tests > >> make[3]: Entering directory `/usr/src/packages/BUILD/R-2.6.2/tests' > >> running code in 'reg-tests-1.R' ... OK > >> running code in 'reg-tests-2.R' ... OK > >> comparing 'reg-tests-2.Rout' to './reg-tests-2.Rout.save' ...1318c1318 > >> < 0.1230 -0.7187 > >> --- > >>> 0.1230 0.3170 > >> 1328c1328 > >> < 0.1230 -0.7187 > >> --- > >>> 0.1230 0.3170 > >> 1356c1356 > >> < -0.2364 -0.3594 > >> --- > >>> 0.2814 0.1585 > >> 1366c1366 > >> < 0.62010 -0.55327 -0.15379 0.02127 > >> --- > >>> 0.49064 -0.68273 -0.02433 0.15074 > >> 1375c1375 > >> < 1.17337 0.06683 -0.15379 0.02127 > >> --- > >>> 1.17337 -0.19209 -0.02433 0.15074 > >> make[3]: *** [reg-tests-2.Rout] Error 1 > >> make[3]: Leaving directory `/usr/src/packages/BUILD/R-2.6.2/tests' > >> make[2]: *** [test-Reg] Error 2 > >> make[2]: Leaving directory `/usr/src/packages/BUILD/R-2.6.2/tests' > >> make[1]: *** [test-all-basics] Error 1 > >> make[1]: Leaving directory `/usr/src/packages/BUILD/R-2.6.2/tests' > >> make: *** [check] Error 2 > >> error: Bad exit status from /var/tmp/rpm-tmp.74662 (%build) > >> > >> __ > >> R-devel@r-project.org mailing list > >> https://stat.ethz.ch/mailman/listinfo/r-devel > >> > > > > -- > > Brian D. Ripley, [EMAIL PROTECTED] > > Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/ > > University of Oxford, Tel: +44 1865 272861 (self) > > 1 South Parks Road, +44 1865 272866 (PA) > > Oxford OX1 3TG, UKFax: +44 1865 272595 > > > > -- > Brian D. Ripley, [EMAIL PROTECTED] > Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/ > University of Oxford, Tel: +44 1865 272861 (self) > 1 South Parks Road, +44 1865 272866 (PA) > Oxford OX1 3TG, UKFax: +44 1865 272595 __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
[Rd] R_ParseVector problem: it's cutting off after the decimal point
Dear all,
my aim is to integrate R in an interactive visualisation software
called Bulk Analyzer developed by VrVis (http://www.vrvis.at).
the code:
SEXP e, tmp;
ParseStatus status;
PROTECT(tmp = mkString("x <- c(1.234,-3.45)"));
PrintValue(tmp);
PROTECT(e = R_ParseVector(tmp, -1, &status, R_NilValue));
PrintValue(e);
UNPROTECT(2);
produces the following output:
[1] "x <- c(1.234,-3.45)"
expression(x <- c(1, -3))
I'm using MS Visual C++. The code works fine in a small test project
with the same R startup settings (attached) as in the Bulk Analyzer
project.
Is there any setting or condition for this parser behaviour?
Kind regards,
Roland N. Boubela
> sessionInfo()
R version 2.6.2 (2008-02-08)
i386-pc-mingw32
locale:
German_Austria.1252
attached base packages:
[1] stats graphics grDevices utils datasets methods base
// init R
Rf_initEmbeddedR(0, NULL);
Rstart Rp = &rp;
R_DefParams(Rp);
// root of the R installation
Rp->rhome = get_R_HOME();
// root of the users directory
Rp->home = getRUser();
// UImode: RGui, RTerm, LinkDLL
Rp->CharacterMode = LinkDLL;
// function which reads input for R from a console
Rp->ReadConsole = myReadConsole;
// function which writes output from R to a console
Rp->WriteConsole = NULL;
Rp->WriteConsoleEx = myWriteConsoleEx;
// called during i/o, eval, graphics in ProcessEvents
Rp->CallBack = myCallBack;
// handles messages for the user from R
Rp->ShowMessage = myShowconcise Message;
// user interaction: a user is asked a question an should answer
with YesNoCancel (maybe dialog)
Rp->YesNoCancel = myYesNoCancel;
// on R state changed: busy(=1)/idle(=0)
Rp->Busy = myBusy;
Rp->home = getRUser();
// R-Settings...
Rp->R_Quiet= (Rboolean)FALSE;
Rp->R_Interactive= (Rboolean)TRUE;
Rp->RestoreAction= SA_RESTORE;
Rp->SaveAction= SA_NOSAVE;
// our parameters
R_SetParams(Rp);
R_set_command_line_arguments(0, NULL);
GA_initapp(0, NULL);
readconsolecfg();
/// run_Rmainloop() starting in a different thread
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
[Rd] code optimization,analysis
Hi, After having used S+ in commercial setting and then R in non-commerical (but still supporting my commercial interests), I would like to give back, if possible. Where I can contribute: 1. SCAN / PREVENT Stability and Threat analysis (big open source presence supported by coverity) 2. Intel Threaded Building Blocks library support to leverage multicore CPUs (plan now for 8-way) 3. VTune analysis/cache optimization for identification of loop unrolls, etc. All this would be on linux (Fed8 at present), except I do have an Intel compiler on Win if it was needed, would need to find fortran compiler, make lots of adjustments, I suspect. I've already started on #3 and would only take on #1 if someone in the R project blessed, because David Miller at coverity likes to support the broader projects over lone-cowboy stuff. P.S. What an easy build, as far as open src goes, your build person deserves credit. I'm currently setting up to compare gcc to intel compiler on the R base and rerunning the timing benchmarks. Thanks. M Jacobson http://www.ihaveyoga.com ph. 415.515. Be a better friend, newshound, and __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] cut.Date and cut.POSIXt problem
Hi all,
Apologies for the delay in my engaging in this thread. I was traveling
this week.
The problem that Gabor raised was caused by the patch that I submitted
to fix a problem with the referenced functions when using 'months' and
'years' as the interval. The prior versions were problematic:
https://stat.ethz.ch/pipermail/r-devel/2008-January/048004.html
The patch fixed the error, but since I used hist.Date() as the reference
model and did not note the subtle difference in cut.Date() relative to
specifying the breaks increment value, this functionality was lost when
the same modification was made to the code in cut.Date().
Roger's patch helps, but does not totally remedy the situation. One also
needs to modify the method used for specifying the max value 'end' for
the breaks in order to include the max 'x' Date value in the result.
Hence, I am attaching proposed patches against R-devel for
base:::dates.R and base:::datetime.R.
I am also attaching a patch for tests:::reg-tests-1.R to add a check for
this situation to the regression tests that were also added subsequent
to that prior set of patches that I had submitted.
If perhaps Roger and Gabor could so some testing on these patches before
they are considered for inclusion into the R-devel tree, it would be
helpful to check to see if I have missed something else here.
Thanks for raising this issue.
Regards,
Marc Schwartz
Roger D. Peng wrote:
Seems changes in r44116 force the interval to be single months (or
years) instead of whatever the user specified. I think the attached
patches correct this.
Interestingly, 'cut' and 'seq' allow for the 'breaks' specification to
be something like "3 months" but the documentation for 'hist' does not
allow for this type of specification.
-roger
Gabor Grothendieck wrote:
cut.Date and cut.POSIXt indicate that the breaks argument
can be an integer followed by a space followed by "year", etc.
but it seems the integer is ignored.
For example, I assume that breaks = "3 months" is supposed
to cut it into quarters but, in fact, it cuts it into months as if
3 had not been there.
d <- seq(Sys.Date(), length = 12, by = "month")
cut(d, "3 months")
[1] 2008-03-01 2008-04-01 2008-05-01 2008-06-01 2008-07-01 2008-08-01
2008-09-01 2008-10-01 2008-11-01 2008-12-01 2009-01-01 2009-02-01
Levels: 2008-03-01 2008-04-01 2008-05-01 2008-06-01 2008-07-01
2008-08-01 2008-09-01 2008-10-01 2008-11-01 2008-12-01 2009-01-01
2009-02-01
cut(as.POSIXct(d), "3 months")
[1] 2008-03-01 2008-04-01 2008-05-01 2008-06-01 2008-07-01 2008-08-01
2008-09-01 2008-10-01 2008-11-01 2008-12-01 2009-01-01 2009-02-01
Levels: 2008-03-01 2008-04-01 2008-05-01 2008-06-01 2008-07-01
2008-08-01 2008-09-01 2008-10-01 2008-11-01 2008-12-01 2009-01-01
2009-02-01
cut(as.POSIXlt(d), "3 months")
[1] 2008-03-01 2008-04-01 2008-05-01 2008-06-01 2008-07-01 2008-08-01
2008-09-01 2008-10-01 2008-11-01 2008-12-01 2009-01-01 2009-02-01
Levels: 2008-03-01 2008-04-01 2008-05-01 2008-06-01 2008-07-01
2008-08-01 2008-09-01 2008-10-01 2008-11-01 2008-12-01 2009-01-01
2009-02-01
--- datesORIG.R 2008-03-20 14:25:13.0 -0500
+++ dates.R 2008-03-20 14:38:21.0 -0500
@@ -322,17 +322,19 @@
if(valid == 3) {
start$mday <- 1
end <- as.POSIXlt(max(x, na.rm = TRUE))
-end <- as.POSIXlt(end + (31 * 86400))
+step <- ifelse(length(by2) == 2, as.integer(by2[1]), 1)
+end <- as.POSIXlt(end + (31 * step * 86400))
end$mday <- 1
-breaks <- as.Date(seq(start, end, "months"))
+breaks <- as.Date(seq(start, end, breaks))
} else if(valid == 4) {
start$mon <- 0
start$mday <- 1
end <- as.POSIXlt(max(x, na.rm = TRUE))
-end <- as.POSIXlt(end + (366 * 86400))
+step <- ifelse(length(by2) == 2, as.integer(by2[1]), 1)
+end <- as.POSIXlt(end + (366 * step * 86400))
end$mon <- 0
end$mday <- 1
-breaks <- as.Date(seq(start, end, "years"))
+breaks <- as.Date(seq(start, end, breaks))
} else {
start <- .Internal(POSIXlt2Date(start))
if (length(by2) == 2) incr <- incr * as.integer(by2[1])
--- datetimeORIG.R 2008-03-20 14:25:20.0 -0500
+++ datetime.R 2008-03-20 15:25:49.0 -0500
@@ -727,17 +727,19 @@
if(valid == 6) {
start$mday <- 1
end <- as.POSIXlt(max(x, na.rm = TRUE))
-end <- as.POSIXlt(end + (31 * 86400))
+step <- ifelse(length(by2) == 2, as.integer(by2[1]), 1)
+end <- as.POSIXlt(end + (31 * step * 86400))
end$mday <- 1
-breaks <- seq(start, end, "months")
+breaks <- seq(start, end, breaks)
} else if(valid == 7) {
start$mon <- 0
start$mday <- 1
end <- as.POSIXlt(max(x, na.rm = TRUE))
-end <- as.POSIXlt(end + (366 * 86400))
+step <- ifelse(length(by2) == 2, as.integer(by2[1]), 1)
+end <- as.POSIXlt(end + (366 * step* 86400))
end$mon
Re: [Rd] R_ParseVector problem: it's cutting off after the decimal point
There are so many things that might be going on here that I at least can't
do more than guess pretty blindly.
First, you are assuming that this is a parser issue, but that's not proved
by your test -- it might be a printing issue, which potentially involves
deparsing. So I would have at least evaluated the expression and printed
the result (via printf, not PrintValue).
I noticed that you did not check the parse status so we don't know if this
parsed completely.
This could well be a locale issue: we don't have a complete program and
you are in a non-English locale. The setting of the locale category
LC_NUMERIC does affect this, and I would not expect embedded R to work
unless that was 'C'.
Here's a test that points to the locale issue:
> Sys.setlocale("LC_NUMERIC", "german")
[1] "German_Germany.1252"
Warning message:
In Sys.setlocale("LC_NUMERIC", "german") :
setting 'LC_NUMERIC' may cause R to function strangely
> foo <- parse(text="x <- c(1.234,-3.45)")
> attributes(foo) <- NULL
> foo
expression(x <- c(1, -3))
So my guess it is an error in the environment you are using for embedded
R.
On Wed, 19 Mar 2008, Roland Boubela wrote:
> Dear all,
>
> my aim is to integrate R in an interactive visualisation software
> called Bulk Analyzer developed by VrVis (http://www.vrvis.at).
>
> the code:
>
> SEXP e, tmp;
> ParseStatus status;
>
> PROTECT(tmp = mkString("x <- c(1.234,-3.45)"));
> PrintValue(tmp);
> PROTECT(e = R_ParseVector(tmp, -1, &status, R_NilValue));
> PrintValue(e);
> UNPROTECT(2);
>
> produces the following output:
>
> [1] "x <- c(1.234,-3.45)"
>
> expression(x <- c(1, -3))
>
>
> I'm using MS Visual C++. The code works fine in a small test project
> with the same R startup settings (attached) as in the Bulk Analyzer
> project.
> Is there any setting or condition for this parser behaviour?
>
> Kind regards,
> Roland N. Boubela
>
>
>> sessionInfo()
> R version 2.6.2 (2008-02-08)
> i386-pc-mingw32
>
> locale:
> German_Austria.1252
>
> attached base packages:
> [1] stats graphics grDevices utils datasets methods base
>
>// init R
>Rf_initEmbeddedR(0, NULL);
>
>Rstart Rp = &rp;
>R_DefParams(Rp);
>
>// root of the R installation
>Rp->rhome = get_R_HOME();
>
>// root of the users directory
>Rp->home = getRUser();
>
>// UImode: RGui, RTerm, LinkDLL
>Rp->CharacterMode = LinkDLL;
>
>// function which reads input for R from a console
>Rp->ReadConsole = myReadConsole;
>
>// function which writes output from R to a console
>Rp->WriteConsole = NULL;
>Rp->WriteConsoleEx = myWriteConsoleEx;
>
>// called during i/o, eval, graphics in ProcessEvents
>Rp->CallBack = myCallBack;
>
>// handles messages for the user from R
>Rp->ShowMessage = myShowconcise Message;
>
>// user interaction: a user is asked a question an should answer
> with YesNoCancel (maybe dialog)
>Rp->YesNoCancel = myYesNoCancel;
>
>// on R state changed: busy(=1)/idle(=0)
>Rp->Busy = myBusy;
>
>Rp->home = getRUser();
>
>// R-Settings...
>Rp->R_Quiet= (Rboolean)FALSE;
>Rp->R_Interactive= (Rboolean)TRUE;
>Rp->RestoreAction= SA_RESTORE;
>Rp->SaveAction= SA_NOSAVE;
>
>// our parameters
>R_SetParams(Rp);
>R_set_command_line_arguments(0, NULL);
>
>GA_initapp(0, NULL);
>readconsolecfg();
>
>/// run_Rmainloop() starting in a different thread
>
> __
> R-devel@r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel
>
--
Brian D. Ripley, [EMAIL PROTECTED]
Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/
University of Oxford, Tel: +44 1865 272861 (self)
1 South Parks Road, +44 1865 272866 (PA)
Oxford OX1 3TG, UKFax: +44 1865 272595
__
R-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
[Rd] Installation error message with 2.6.2-win32 (PR#11007)
Full_Name: Kevin Zembower Version: 2.6.2 OS: Windows XP Pro, ver 2002 SP2 Submission from: (NULL) (162.129.44.50) Noticed the following errors and artifacts when trying to install R 2.6.2-win32.exe on my Windows XP Pro version 2002 Service Pack 2 workstation. I first attempted to install it as the 'Administrator'. More than once, I got the message, "Internal Error: Fialed to expand shell folder constant 'userdocs'". Note that this is a networked PC at my work. I am not certain that the Administrator has a home directory or that it is accessible. I noticed that there was no entry for R itself in the Start->Programs->R folder, just ones for Help and Uninstall. Next, I tried installing the same file as a normal user on my system. I turned off all registry entries, as I thought this was the problem the normal user failed to install R the last time I tried this. This time, I got the error "Setup was unable to create the directory "Y:\Start Menu\Programs\R" Error 5: Access is denied." However, it did create Quick Launch and Desktop icons, that seem to work correctly. R 2.6.2 seems to run correctly. As the error message indicates, however, there is still no R entry itself in Start->Programs->R. Please let me know if I can provide more information. -Kevin Zembower __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] code optimization,analysis
On 20/03/2008 7:03 AM, m jacobson wrote: > Hi, > > After having used S+ in commercial setting and then R > in non-commerical (but still supporting my commercial > interests), I would like to give back, if possible. > > Where I can contribute: > 1. SCAN / PREVENT Stability and Threat analysis (big > open source presence supported by coverity) > 2. Intel Threaded Building Blocks library support to > leverage multicore CPUs (plan now for 8-way) > 3. VTune analysis/cache optimization for > identification of loop unrolls, etc. > > All this would be on linux (Fed8 at present), except I > do have an Intel compiler on Win if it was needed, > would need to find fortran compiler, make lots of > adjustments, I suspect. > > I've already started on #3 and would only take on #1 > if someone in the R project blessed, because David > Miller at coverity likes to support the broader > projects over lone-cowboy stuff. You didn't mention this, but putting together (and maintaining) a tool chain for 64 bit Windows builds would be appreciated. > P.S. What an easy build, as far as open src goes, your > build person deserves credit. I'm currently setting > up to compare gcc to intel compiler on the R base and > rerunning the timing benchmarks. That's largely due to Brian Ripley's efforts, but there are many other contributors to it. Duncan Murdoch __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] Roxygen
> No, we want a solution in R. Would it suffice, by the way, to source() a file and introspect upon its objects with ls(), formals(), typeof(), mode(), and the like; or should we formalize, say, a BNF and write the accompanying automaton? __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
