Re: [Rd] cat: ./R/Copy: No such file or directory
Please note that the assmuption is that you can install a pacakge before you check it. R CMD INSTALL often gives fuller diagnostics: I am not sure these would have helped you here, but I would have expected you to have got a 'syntax error' message (which is not 'cryptic': you did have syntax errors in your files). On Sat, 1 Nov 2008, Spencer Graves wrote: Dear Duncan: in line Duncan Murdoch wrote: On 01/11/2008 8:53 PM, Spencer Graves wrote: Dear Ted: Thanks very much. In fact, I did have a file named Copy of create.fourier.basis.R. I deleted it and got an even more cryptic version of the same error message. I ultimately traced this to lines like .working inserted by subversion into a *.R or *.Rd file I was using. I found this by finding a previous *.tar.gz file that passed R CMD check, studying the subversion log of 7 different changes made by 3 people since that one that worked, and reviewing changes made to at least 13 different files in that period until I found the problem. Hi Spencer. If you're using Windows, I really recommend using TortoiseSVN. It'll make it very easy to see conflict markers like the one that caused your trouble. (I don't like their compare utility, but it's easy to plug in a different one. I use Beyond Compare.) I've been using TortoiseSVN for roughly 18 months now, but I had previously not used much of its capabilities. The log and diff files helped narrow the problem from someplace in 400 files to someplace in 13. It would be nice if the R CMD check diagnostics were more clear about things like this. However, even with these problems, it is vastly superior to the anarchy that organizes the software development efforts of many people I know. I have friends and colleagues developing software in other languages with NOTHING like this, asking me to check their work. I'm refining lectures on the virtues of R CMD check and pushing others adopt something similar for perl scripts, Matlab code, etc. I'm eager to help develop such systems, and I resist efforts to have me repeat tests I developed months ago and should be automated. Thanks for your reply -- and for your very valuable work in helping to develop the current R CMD check system. Best Wishes, Spencer If you're not on Windows, there are probably similar things available, but I don't know what they are. Duncan Murdoch Best Wishes, Spencer (Ted Harding) wrote: Just guessing here, but it looks as though an attempt has been made to execute the following command: cat ./R/Copy of create.fourier.basis.R This may have arisen because there is indeed a file named Copy of create.fourier.basis.R but the command was issued to the OS without quotes; or (perhaps less likely, because of the presence of the ./R/) the line Copy of create.fourier.basis.R was intended as a comment in the file, but somehow got read as a substantive part of the code. Good luck! Ted. On 01-Nov-08 15:40:33, Spencer Graves wrote: Hello: What do you recommend I do to get past cryptic error messages from R CMD check, complaining No such file or directory? The package is under SVN control, and I reverted to a version that passed R CMD check, without eliminating the error. The 00install.out file is short and bitter: cat: ./R/Copy: No such file or directory cat: of: No such file or directory cat: create.fourier.basis.R: No such file or directory make: *** [Rcode0] Error 1 -- Making package fda adding build stamp to DESCRIPTION installing NAMESPACE file and metadata make[2]: *** No rule to make target `R/Copy', needed by `D:/spencerg/statmtds/splines/fda/RForge/fda/fda.Rcheck/fda/R/fda'. Stop. make[1]: *** [all] Error 2 make: *** [pkg-fda] Error 2 *** Installation of fda failed *** Removing 'D:/spencerg/statmtds/splines/fda/RForge/fda/fda.Rcheck/fda' Thanks for any suggestions. I have a *.tar.gz file that I hope may not have this problem, but I'm not sure. Thanks, Spencer __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel E-Mail: (Ted Harding) [EMAIL PROTECTED] Fax-to-email: +44 (0)870 094 0861 Date: 01-Nov-08 Time: 16:19:56 -- XFMail -- __ 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 -- 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
[Rd] Emacs tags for R code
Hi, I have put up details of a new tool for tagging R code at http://dsarkar.fhcrc.org/rtags/rtags.html that Emacs/ESS users may be interested in. It should be possible to extend this for other editors as well. Comments and feedback welcome. -Deepayan __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
[Rd] Wishlist: pass args from demo() to source()
Currently demo calls source with a hard coded max.deparse.length = 250 so you can't really see the demo properly in some cases. Note the [TRUNCATED] below. (1) It would be nice if demo passed max.deparse.length (and other args to source). (2) Also a larger max.deparse.length default would be nice to make it less likely one would have to set it in the first place. R.version.string # Vista [1] R version 2.8.0 Patched (2008-10-21 r46766) demo(gsubfn-si) demo(gsubfn-si) ~ Type Return to start : # given number possibly followed by SI letter (e.g. 32.5k where k means 1000) # replace letter with e followed by appropriate digits. # (see formatEng2R by Hans-Joerg Bibiko in the R Wiki) conv - list(y = e-24, z = e-21, a = e-18, f [TRUNCATED] gsubfn(.$, conv, c(19, 32.5M)) [1] 19 32.5e6 __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] Wishlist: pass args from demo() to source()
On 02/11/2008 9:33 AM, Gabor Grothendieck wrote: Currently demo calls source with a hard coded max.deparse.length = 250 so you can't really see the demo properly in some cases. Note the [TRUNCATED] below. (1) It would be nice if demo passed max.deparse.length (and other args to source). (2) Also a larger max.deparse.length default would be nice to make it less likely one would have to set it in the first place. I can see increasing the default, but it doesn't really seem like a good design to give extra args to demo(). Would users ever know about them? So to me the choice would be to set a large max.deparse.length in the call to source() from demo(). Just for fun, I tracked down the revision where the 250 value was introduced: it was in r3045 in December 1998 on https://svn.r-project.org/R/branches/R-0-63-patches/src/library/base/R/source.R, where the max was decreased from 1 to 250. (The demo() function was in source.R in those days.) And the 1 was added in r354 in December 1997 to https://svn.r-project.org/R/trunk/src/library/base/R/source, to remove the TRUNCATED message. Martin, is there any chance you remember what problem the 1 caused? If those problems still exist, I'd add an arg to demo with 250 as the default, but if not, I'd say putting it back as 1 would be reasonable. Duncan Murdoch R.version.string # Vista [1] R version 2.8.0 Patched (2008-10-21 r46766) demo(gsubfn-si) demo(gsubfn-si) ~ Type Return to start : # given number possibly followed by SI letter (e.g. 32.5k where k means 1000) # replace letter with e followed by appropriate digits. # (see formatEng2R by Hans-Joerg Bibiko in the R Wiki) conv - list(y = e-24, z = e-21, a = e-18, f [TRUNCATED] gsubfn(.$, conv, c(19, 32.5M)) [1] 19 32.5e6 __ 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] cat: ./R/Copy: No such file or directory
Dear Prof. Ripley: Thanks for the suggestion. I've used some of those diagnostics, and I think I knew that I could run R CMD install --build without having passed R CMD check, but I hadn't done it before and didn't think of it. I will remember it for the future. Best Wishes, Spencer Prof Brian Ripley wrote: Please note that the assmuption is that you can install a pacakge before you check it. R CMD INSTALL often gives fuller diagnostics: I am not sure these would have helped you here, but I would have expected you to have got a 'syntax error' message (which is not 'cryptic': you did have syntax errors in your files). On Sat, 1 Nov 2008, Spencer Graves wrote: Dear Duncan: in line Duncan Murdoch wrote: On 01/11/2008 8:53 PM, Spencer Graves wrote: Dear Ted: Thanks very much. In fact, I did have a file named Copy of create.fourier.basis.R. I deleted it and got an even more cryptic version of the same error message. I ultimately traced this to lines like .working inserted by subversion into a *.R or *.Rd file I was using. I found this by finding a previous *.tar.gz file that passed R CMD check, studying the subversion log of 7 different changes made by 3 people since that one that worked, and reviewing changes made to at least 13 different files in that period until I found the problem. Hi Spencer. If you're using Windows, I really recommend using TortoiseSVN. It'll make it very easy to see conflict markers like the one that caused your trouble. (I don't like their compare utility, but it's easy to plug in a different one. I use Beyond Compare.) I've been using TortoiseSVN for roughly 18 months now, but I had previously not used much of its capabilities. The log and diff files helped narrow the problem from someplace in 400 files to someplace in 13. It would be nice if the R CMD check diagnostics were more clear about things like this. However, even with these problems, it is vastly superior to the anarchy that organizes the software development efforts of many people I know. I have friends and colleagues developing software in other languages with NOTHING like this, asking me to check their work. I'm refining lectures on the virtues of R CMD check and pushing others adopt something similar for perl scripts, Matlab code, etc. I'm eager to help develop such systems, and I resist efforts to have me repeat tests I developed months ago and should be automated. Thanks for your reply -- and for your very valuable work in helping to develop the current R CMD check system. Best Wishes, Spencer If you're not on Windows, there are probably similar things available, but I don't know what they are. Duncan Murdoch Best Wishes, Spencer (Ted Harding) wrote: Just guessing here, but it looks as though an attempt has been made to execute the following command: cat ./R/Copy of create.fourier.basis.R This may have arisen because there is indeed a file named Copy of create.fourier.basis.R but the command was issued to the OS without quotes; or (perhaps less likely, because of the presence of the ./R/) the line Copy of create.fourier.basis.R was intended as a comment in the file, but somehow got read as a substantive part of the code. Good luck! Ted. On 01-Nov-08 15:40:33, Spencer Graves wrote: Hello: What do you recommend I do to get past cryptic error messages from R CMD check, complaining No such file or directory? The package is under SVN control, and I reverted to a version that passed R CMD check, without eliminating the error. The 00install.out file is short and bitter: cat: ./R/Copy: No such file or directory cat: of: No such file or directory cat: create.fourier.basis.R: No such file or directory make: *** [Rcode0] Error 1 -- Making package fda adding build stamp to DESCRIPTION installing NAMESPACE file and metadata make[2]: *** No rule to make target `R/Copy', needed by `D:/spencerg/statmtds/splines/fda/RForge/fda/fda.Rcheck/fda/R/fda'. Stop. make[1]: *** [all] Error 2 make: *** [pkg-fda] Error 2 *** Installation of fda failed *** Removing 'D:/spencerg/statmtds/splines/fda/RForge/fda/fda.Rcheck/fda' Thanks for any suggestions. I have a *.tar.gz file that I hope may not have this problem, but I'm not sure. Thanks, Spencer __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel E-Mail: (Ted Harding) [EMAIL PROTECTED] Fax-to-email: +44 (0)870 094 0861 Date: 01-Nov-08 Time: 16:19:56 -- XFMail -- __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel __
[Rd] R CMD check with Matlab and perl?
Hi, All: How might one test Matlab and perl code during R CMD check? I ask for two reasons: First, the fda package exists in both R and Matlab, and it would help if we could run test examples in Matlab as part of R CMD check. I tried R.matlab, but could not get it to work in that context. Second, I have professional colleagues who write code in Matlab and perl, but have no unit or regression testing. If I could test Matlab and perl from within R CMD check examples, I think I could get them to use it, because it would produce more trustworthy software [the Prime Directive of Chambers (2008) Software for Data Analysis (Springer)] AND give them documentation at the same time -- while simultaneously reducing the cost of producing and maintaining code to given specifications. Thanks, Spencer __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
[Rd] Thanks for R-2.8.0 still on W2000
Hello, This little post is just to thank Prof. Ripley and the R team for still maintaining R runable on Windows 2000. I installed R-2.8.0 on Windows 2000 and used it for several computation hours and everything seems to work fine. Many thanks Vincent __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] R CMD check with Matlab and perl?
Spencer Graves wrote: Hi, All: How might one test Matlab and perl code during R CMD check? I ask for two reasons: First, the fda package exists in both R and Matlab, and it would help if we could run test examples in Matlab as part of R CMD check. I tried R.matlab, but could not get it to work in that context. Can you run the tests from the command line? If so, you should be able to call them from system() in a script in the tests directory, or in an example section of an Rd file. Duncan Murdoch Second, I have professional colleagues who write code in Matlab and perl, but have no unit or regression testing. If I could test Matlab and perl from within R CMD check examples, I think I could get them to use it, because it would produce more trustworthy software [the Prime Directive of Chambers (2008) Software for Data Analysis (Springer)] AND give them documentation at the same time -- while simultaneously reducing the cost of producing and maintaining code to given specifications. Thanks, Spencer __ 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
