I've been implementing a wrapper to the 2011 Fortran version of L-BFGS-B. In optim(), R uses a C translation of a Fortran version (the version number does not appear to be documented by the original authors). The authors of the original Fortran code have updated it and published the reasons in ACM TOMS due to inefficiencies and a bug. In running the checks on the resulting package (which is on R-forge under the optimizer project), I'm getting a number of warning messages of the type Warning in file.copy(file.path(.Library, pkg, "DESCRIPTION"), pd) : problem copying /usr/lib/R/library/mgcv/DESCRIPTION to /tmp/Rtmp0kkeHo/RLIBS_1214765d1c5f/mgcv/DESCRIPTION: No such file or directory which reference DESCRIPTIONs for a number of packages other than the one being checked -- here mgcv -- and which are not referenced in my package as far as I can determine. Possibly unrelated, when I run the code on a problem, it works for one run, then gives a NAMESPACE error and failure on the second try. Apart from this, checks and unit tests appear to work correctly. Does anyone have pointers where I might find some ideas on the origin of the issue(s)? I suspect the warning messages are not particularly indicative of the source of the warnings, but that I have some subtle glitch in the setup and call to the Fortran. I suspect this is not platform dependent, but I'm running Linux Mint 17.1 (ubuntu derivative), and R 3.1.2. Cheers, JN
>>>>> Prof J C Nash (U30A) writes:> I've been implementing a wrapper to the 2011 Fortran version of > L-BFGS-B. In optim(), R uses a C translation of a Fortran version (the > version number does not appear to be documented by the original > authors). The authors of the original Fortran code have updated it and > published the reasons in ACM TOMS due to inefficiencies and a bug.> In running the checks on the resulting package (which is on R-forge > under the optimizer project), I'm getting a number of warning messages > of the type> Warning in file.copy(file.path(.Library, pkg, "DESCRIPTION"), pd) : > problem copying /usr/lib/R/library/mgcv/DESCRIPTION to > /tmp/Rtmp0kkeHo/RLIBS_1214765d1c5f/mgcv/DESCRIPTION: No such file or > directory> which reference DESCRIPTIONs for a number of packages other than the one > being checked -- here mgcv -- and which are not referenced in my package > as far as I can determine.> Possibly unrelated, when I run the code on a problem, it works for one > run, then gives a NAMESPACE error and failure on the second try. Apart > from this, checks and unit tests appear to work correctly.> Does anyone have pointers where I might find some ideas on the origin of > the issue(s)? I suspect the warning messages are not particularly > indicative of the source of the warnings, but that I have some subtle > glitch in the setup and call to the Fortran.> I suspect this is not platform dependent, but I'm running Linux Mint > 17.1 (ubuntu derivative), and R 3.1.2.John: maybe you did not install the recommended packages? (On Debian, the corresponding package would be r-recommended.) Best -k> Cheers, JN> ______________________________________________ > R-devel at r-project.org mailing list > https://stat.ethz.ch/mailman/listinfo/r-devel
On 18 January 2015 at 09:03, Prof J C Nash (U30A) wrote: | I've been implementing a wrapper to the 2011 Fortran version of | L-BFGS-B. In optim(), R uses a C translation of a Fortran version (the | version number does not appear to be documented by the original | authors). The authors of the original Fortran code have updated it and | published the reasons in ACM TOMS due to inefficiencies and a bug. | | In running the checks on the resulting package (which is on R-forge | under the optimizer project), I'm getting a number of warning messages | of the type If you were so kind to share the __name of the subpackage__ you fail to test, one could attempt to help you. | Warning in file.copy(file.path(.Library, pkg, "DESCRIPTION"), pd) : | problem copying /usr/lib/R/library/mgcv/DESCRIPTION to | /tmp/Rtmp0kkeHo/RLIBS_1214765d1c5f/mgcv/DESCRIPTION: No such file or | directory | | which reference DESCRIPTIONs for a number of packages other than the one | being checked -- here mgcv -- and which are not referenced in my package | as far as I can determine. I've seen that when R tries to be too clever by half -- somehow .libPaths() ends up being partial. Following the high-level decision by Hornik, Leisch and Eddelbuettel made circa 2003 in a pub in Vienna, the Debian packages use three entries, and if you have recommended packages installed via apt, R may now pretend they don't exist. I could work on an alternate setup via Github and Travis but as you managed to make this non-reproducible I cannot actually try that ... Dirk -- http://dirk.eddelbuettel.com | @eddelbuettel | edd at debian.org
Kurt pointed to the issue. Thanks. I did install r-recommended, but it seems something went wrong at some point. A reinstall got rid of the warnings. Thanks to Dirk for his offer of help. Now I'm still getting a namespace issue on the second run of an optimization problem. However, I think I need to do some more digging to narrow down where this issue is lurking. It may be some local matter, as with the r-recommended links failing. Best, JN On 15-01-18 09:27 AM, Kurt Hornik wrote:>>>>>> Prof J C Nash (U30A) writes: > >> I've been implementing a wrapper to the 2011 Fortran version of >> L-BFGS-B. In optim(), R uses a C translation of a Fortran version (the >> version number does not appear to be documented by the original >> authors). The authors of the original Fortran code have updated it and >> published the reasons in ACM TOMS due to inefficiencies and a bug. > >> In running the checks on the resulting package (which is on R-forge >> under the optimizer project), I'm getting a number of warning messages >> of the type > >> Warning in file.copy(file.path(.Library, pkg, "DESCRIPTION"), pd) : >> problem copying /usr/lib/R/library/mgcv/DESCRIPTION to >> /tmp/Rtmp0kkeHo/RLIBS_1214765d1c5f/mgcv/DESCRIPTION: No such file or >> directory > >> which reference DESCRIPTIONs for a number of packages other than the one >> being checked -- here mgcv -- and which are not referenced in my package >> as far as I can determine. > >> Possibly unrelated, when I run the code on a problem, it works for one >> run, then gives a NAMESPACE error and failure on the second try. Apart >> from this, checks and unit tests appear to work correctly. > >> Does anyone have pointers where I might find some ideas on the origin of >> the issue(s)? I suspect the warning messages are not particularly >> indicative of the source of the warnings, but that I have some subtle >> glitch in the setup and call to the Fortran. > >> I suspect this is not platform dependent, but I'm running Linux Mint >> 17.1 (ubuntu derivative), and R 3.1.2. > > John: maybe you did not install the recommended packages? > (On Debian, the corresponding package would be r-recommended.) > > Best > -k > >> Cheers, JN > >> ______________________________________________ >> R-devel at r-project.org mailing list >> https://stat.ethz.ch/mailman/listinfo/r-devel