Dear Duncan, dear Uwe,
Just now I have re-run everything, and today xps.Rnw can be converted to
a vignette w/o any problems using:
a, buildVignettes("xps", dir="/Volumes/CoreData/CRAN/xps", quiet=F)
b, R CMD Sweave xps.Rnw
In both cases the vignette xps.pdf is created (maybe my Mac did not like
to work during eastern holidays).
However, one issue remains:
"R64 CMD check xps_1.13.1.tar.gz" no longer creates any pdf files for
the vignettes.
Best regards
Christian
On 4/25/11 9:31 PM, Duncan Murdoch wrote:
On 25/04/2011 3:16 PM, cstrato wrote:
Thank you.
My problem seems to be that at the moment the problem can be seen only
on my Mac, since e.g. the Bioconductor servers have no problems creating
the vignettes.
Then you are definitely the one in the best position to diagnose the
problem. Use the usual approach: simplify it by cutting out everything
that looks unrelated. Verify that the problem still exists, then cut
some more. Eventually you'll have isolated the error to a particular
small snippet of code, and then you can add print() statements, or use
trace(), or do whatever is necessary to see what's so special about your
system.
I suspect it will turn out to be an assumption in the code that is not
true on your system.
If the assumption is being made by code you wrote, then fix it. If it's
being made by R, let us know.
Duncan Murdoch
Best regards
Christian
On 4/25/11 8:55 PM, Duncan Murdoch wrote:
cstrato wrote:
Dear Duncan,
Thank you for your example, however it is different since it does not
use x and y. What about print(x+y)?
Try it.
Sorry, I do not believe that there is a bug in my code, since:
1, it worked in all versions from R starting with R-2.6.0 till
R-2.12.2.
2, the identical code works in the examples
3, this code (or a similar code) is the starting code which all users
of xps have to use, and there was never a problem.
This might be a problem in R, or might be a problem in your code. As far
as I know, it has only shown up in your code, so I'd guess that's where
the problem is. In any case, you're the one in the best position to
isolate it and debug it.
If it turns out to be a problem in R, put together an example
illustrating the problem that doesn't involve your code, and I'll take a
look.
Duncan Murdoch
Maybe the reason could be that my code has to import
- the CEL-files from the package dir
- the file SchemeTest3.root from the package dir
??
Best regards
Christian
On 4/25/11 8:00 PM, Duncan Murdoch wrote:
cstrato wrote:
Dear Uwe,
Your suggestion to look at the Sweave manual helped me to solve the
problem. It seems that in R-2.13.0 every chunk can use the code from
the chunk before but not from an earlier chunk.
I'm either misreading what you wrote, or it's wrong. If I have this
in a
Sweave file:
<<>>=
x<- 1
@
<<>>=
y<- 2
@
<<>>=
print(x)
@
I will see the value of x getting printed, even though it came from
two
chunks earlier.
I think Uwe is right: there is some bug in the code you're running.
Sweave isn't the problem.
Duncan Murdoch
Concretely, the following does not work since chunk 5 needs the code
from chunk 3 and 4:
###################################################
### chunk number 3:
###################################################
#line 126 "xps.Rnw"
celdir<- file.path(.path.package("xps"), "raw")
###################################################
### chunk number 4:
###################################################
#line 132 "xps.Rnw"
scheme.test3<- root.scheme(file.path(.path.package("xps"), "schemes",
"SchemeTest3.root"))
###################################################
### chunk number 5:
###################################################
#line 137 "xps.Rnw"
celfiles<- c("TestA1.CEL","TestA2.CEL")
data.test3<- import.data(scheme.test3, "tmpdt_DataTest3",
celdir=celdir, celfiles=celfiles, verbose=FALSE)
However, when I add "celdir" to chunk 5 then everything works since
now chunk 5 needs only the code from chunk 4 but not from chunk 3:
###################################################
### chunk number 5:
###################################################
#line 137 "xps.Rnw"
celdir<- file.path(.path.package("xps"), "raw")
celfiles<- c("TestA1.CEL","TestA2.CEL")
data.test3<- import.data(scheme.test3, "tmpdt_DataTest3",
celdir=celdir, celfiles=celfiles, verbose=FALSE)
Now buildVignettes() is able to create the vignettes, however R CMD
check still does not build the vignettes.
Yes, I get a Warning in both cases:
* checking package vignettes in 'inst/doc' ... WARNING
Package vignettes without corresponding PDF: ........
However, with R-2.12.2 the following lines are added:
/Volumes/CoreData/CRAN/xps.Rcheck/00_pkg_src/xps/inst/doc/APTvsXPS.Rnw
/Volumes/CoreData/CRAN/xps.Rcheck/00_pkg_src/xps/inst/doc/xps.Rnw
/Volumes/CoreData/CRAN/xps.Rcheck/00_pkg_src/xps/inst/doc/xpsClasses.Rnw
/Volumes/CoreData/CRAN/xps.Rcheck/00_pkg_src/xps/inst/doc/xpsPreprocess.Rnw
and in xps.Rcheck the subdirectory "inst/doc" will be created which
contains the vignette data such as xps.Rnw, but also xps.tex and
xps.pdf.
In contrast, R-2.13.0 does not create the subdirectory "inst/doc" and
no vignettes are built.
One more issue:
In contrast to my former believe R CMD INSTALL does not build the
vignettes, neither in R-2.12.2 nor in R-2.13.0. I have to run
buildVignettes() after installation. Is this the usual case?
Best regards
Christian
On 4/25/11 4:00 PM, Uwe Ligges wrote:
On 24.04.2011 23:10, cstrato wrote:
Dear Uwe,
On 4/24/11 10:37 PM, Uwe Ligges wrote:
On 24.04.2011 20:59, cstrato wrote:
Dear Uwe,
Thank you for your reply.
ad 2, Yes, i know that "xps-manual.pdf" is the collection of help
pages,
I have mentioned it only to show that creating pdf-files does
work
for
R-2.13.0.
ad 1, Could it be that this is a Mac-specific problem since I see
it on
both my old MacBook Pro and my new Mac Mini.
Have you tried on any other OS? I did not since installing root is
a bit
too much effort.
No, until now I did not try another OS, but I will, since xps
has to
work on all three OSes.
Using R CMD check with R-2.12.2 I get:
I thought we are talking about R-2.13.0?
I showed you the output of R-2.12.2 first and then the output of
R-2.13.0, so that you can see that on the same machine with the
same
file it works with one version but not the other.
But you got a Warning in both cases:
* checking package vignettes in 'inst/doc' ... WARNING
Package vignettes without corresponding PDF: ........
Where is the R CMD build output (since R CMD build is supposed to
prepare the vignette).
$ R64 CMD check xps_1.13.1.tar.gz
* using log directory '/Volumes/CoreData/CRAN/xps.Rcheck'
* using R version 2.12.2 (2011-02-25)
* using platform: x86_64-apple-darwin9.8.0 (64-bit)
* using session charset: ASCII
* checking for file 'xps/DESCRIPTION' ... OK
* this is package 'xps' version '1.13.1'
...
...
* checking package vignettes in 'inst/doc' ... WARNING
Package vignettes without corresponding PDF:
/Volumes/CoreData/CRAN/xps.Rcheck/00_pkg_src/xps/inst/doc/APTvsXPS.Rnw
/Volumes/CoreData/CRAN/xps.Rcheck/00_pkg_src/xps/inst/doc/xps.Rnw
/Volumes/CoreData/CRAN/xps.Rcheck/00_pkg_src/xps/inst/doc/xpsClasses.Rnw
/Volumes/CoreData/CRAN/xps.Rcheck/00_pkg_src/xps/inst/doc/xpsPreprocess.Rnw
* checking PDF version of manual ... OK
Using "RSwitch.app" I switch to R-2.13.0, but now I get:
$ R64 CMD check xps_1.13.1.tar.gz
* using log directory '/Volumes/CoreData/CRAN/xps.Rcheck'
* using R version 2.13.0 (2011-04-13)
* using platform: x86_64-apple-darwin9.8.0 (64-bit)
* using session charset: ASCII
* checking for file 'xps/DESCRIPTION' ... OK
* this is package 'xps' version '1.13.1'
...
...
* checking package vignettes in 'inst/doc' ... WARNING
Package vignette(s) without corresponding PDF:
APTvsXPS.Rnw
xps.Rnw
xpsClasses.Rnw
xpsPreprocess.Rnw
* checking running R code from vignettes ... OK
* checking re-building of vignettes ... OK
* checking PDF version of manual ... OK
I must admit that I have never built the vignettes manually,
and I
cannot find a hint how I can do it from the command line. Is this
possible?
Is it possible to build the vignettes from the command line?
R --help suggests there is
R CMD Sweave .....
However, building the vignettes within R-2.13.0 I get:
library(tools)
buildVignettes("xps", dir="/Volumes/CoreData/CRAN/xps", quiet=F)
Overfull \vbox (21.68121pt too high) has occurred while
\output is
Writing to file xps.tex
Processing code chunks with options ...
1 : term verbatim
2 : echo term hide
3 : echo term verbatim
4 : echo term verbatim
5 : echo term verbatim
SysError in<TFile::TFile>: file
/tmpdt_DataTest3_cel.root/tmpdt_DataTest3_cel_20110424_201301.root
can
not be opened (No such file or directory)
Error: Could not create file</tmpdt_DataTest3_cel.root>
Here are the two important chunks 4 and 5, which work fine with
R-2.12.2
and earlier versions:
###################################################
### chunk number 4:
###################################################
#line 132 "xps.Rnw"
scheme.test3<- root.scheme(file.path(.path.package("xps"),
"schemes",
"SchemeTest3.root"))
###################################################
### chunk number 5:
###################################################
#line 137 "xps.Rnw"
celfiles<- c("TestA1.CEL","TestA2.CEL")
data.test3<- import.data(scheme.test3, "tmpdt_DataTest3",
celdir=celdir, celfiles=celfiles, verbose=FALSE)
However, in R-2.13.0 chunk 5 crashes!!!
It works only when replacing chunk 5 with:
###################################################
### chunk number 5:
###################################################
#line 137 "xps.Rnw"
celfiles<- c("TestA1.CEL","TestA2.CEL")
scheme.test3<- root.scheme(file.path(.path.package("xps"),
"schemes",
"SchemeTest3.root"))
data.test3<- import.data(scheme.test3, "tmpdt_DataTest3",
celdir=celdir, celfiles=celfiles, verbose=FALSE)
As you see R-2.13.0 does no longer remember the result of
chunk 4,
i.e.
"scheme.test3". Now I have to include the line defining
"scheme.test3"
also in chunk 5.
How objects generated in one chunk can be reused later on is
explained
in the Sweave manual.
Do you have any idea for this behavior?
An example, which does work in R-2.13.0 are the following two
chunks:
#################################################
### chunk number 20:
###################################################
#line 300 "xps.Rnw"
library(xps)
scheme.test3<- root.scheme(file.path(.path.package("xps"),
"schemes",
"SchemeTest3.root"))
data.test3<- root.data(scheme.test3,
file.path(.path.package("xps"),"rootdata", "DataTest3_cel.root"))
###################################################
### chunk number 21:
###################################################
#line 318 "xps.Rnw"
data.rma<- rma(data.test3, "tmpdt_Test3RMA", verbose=FALSE)
In summary it is not quite clear to me what has changed in
R-2.13.0 so
that chunk 5 no longer works.
BTW, the error:
SysError in<TFile::TFile>: file
/tmpdt_DataTest3_cel.root/tmpdt_DataTest3_cel_20110424_201301.root
can
not be opened
is not clear to me since "/tmpdt_DataTest3_cel.root/" is
definitively
NOT a directory.
Time to debug what root.scheme is doing with the supplied path.
Sorry, I do not understand what you mean. root.scheme does not
change
the path, otherwise it would not work on all older versions of R.
Right, but since something changed obviously, and you tell us the
path
is not a path, it might be a function used by root.scheme ....
That's
why I said time to debug the code in your vignette!
Do you have any ideas?
Please note that the vignette "xps.Rnw" did work for the last two
years
w/o problem. Furthermore, the Bioconductor servers are able to
build
the
vignette, see:
http://www.bioconductor.org/packages/release/bioc/html/xps.html
See
http://bioconductor.org/checkResults/2.8/bioc-20110225/xps/liverpool-checksrc.html
and find that their Mac Servers are also fine with it. So maybe
it is
your setup that is corrupted?
Maybe, you are right that my setup is corrupted, but on two
independent
machines?
Maybe the download of "R-2.13.0.pkg" is corrupted, but then nothing
should work, or am I wrong?
How can I check if my setup is corrupted?
By debugging the code in your package's vignette.
Uwe Ligges
Best regards
Christian
Best wishes,
Uwe
Best regards
Christian
On 4/24/11 4:12 PM, Uwe Ligges wrote:
On 23.04.2011 21:50, cstrato wrote:
Dear all,
While R CMD check and R CMD INSTALL have always created the
vignettes on
R-2.12.1 or any earlier versions of R, I am no longer able to
build
the
vignettes on R-2.13.0.
Instead R CMD check gives me the following output:
* checking for unstated dependencies in vignettes ... OK
* checking package vignettes in 'inst/doc' ... WARNING
Package vignette(s) without corresponding PDF:
APTvsXPS.Rnw
xps.Rnw
xpsClasses.Rnw
xpsPreprocess.Rnw
* checking running R code from vignettes ... OK
* checking re-building of vignettes ... OK
* checking PDF version of manual ... OK
Does someone know what the reason might be?
No, it does for me for other packages.
Perhaps an error when processing the vignettes? Have you
tried to
build
them manually?
(R64 CMD check --help says that be default rebuild-vignettes is
turned
on.)
Interestingly, R CMD check still creates the file
"xps-manual.pdf".
That is the collection of help pages, unrelated to the vignette.
Uwe Ligges
Here is my sessionInfo:
sessionInfo()
R version 2.13.0 (2011-04-13)
Platform: x86_64-apple-darwin9.8.0/x86_64 (64-bit)
locale:
[1] C
attached base packages:
[1] stats graphics grDevices utils datasets methods base
other attached packages:
[1] xps_1.13.1
loaded via a namespace (and not attached):
[1] Biobase_2.12.1 Biostrings_2.20.0 IRanges_1.10.0
[4] affy_1.30.0 affyPLM_1.28.5 affyio_1.20.0
[7] preprocessCore_1.14.0
Thank you in advance.
Best regards
Christian
_._._._._._._._._._._._._._._._._._
C.h.r.i.s.t.i.a.n S.t.r.a.t.o.w.a
V.i.e.n.n.a A.u.s.t.r.i.a
e.m.a.i.l: cstrato at aon.at
_._._._._._._._._._._._._._._._._._
______________________________________________
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
