[Rd] Compiling error with glmnet and nloptr

jing hua zhao j|nghu@zh@o @end|ng |rom hotm@||@com
Thu Mar 10 16:43:11 CET 2022


Hi Dirk,

Thank for your help -- as it is our university HPC system I might forward your information. Even a recent branch here uses CentOS8 which appears to have the same fate -- a lot of object/executable files compiled at SL6 do not work on CenOS8 which is really a nuisance. In this case, I have a bit luck since it does work in both ways,

>  library(help=RcppNLoptExample)
> testConstrainedProblem("MMA", TRUE)
Found minimum at f(0.33333,0.2963) = 0.54433105 after 11 function and 22 constraint evaluations.
[1] 0.3333333 0.2962963

Best regards,


JIng Hua

________________________________
From: Dirk Eddelbuettel <dirk.eddelbuettel using gmail.com> on behalf of Dirk Eddelbuettel <edd using debian.org>
Sent: 10 March 2022 15:19
To: jing hua zhao <jinghuazhao using hotmail.com>
Cc: R-devel <r-devel using r-project.org>
Subject: Re: [Rd] Compiling error with glmnet and nloptr


Dear jing hua zhao,

You appear to have issues with a particular (pair or single) package; it is
usually better to contact the author / maintainers of the package.  I help a
little with nloptr which had its share of similar issues after its recent as
it (conditionall, if no libnlopr is found) will require cmake, and in a
fairly recentish flavor.  That can pose issues on systems such as yours which
identifies itself per the error message as Scientific Linux

   linux-scientific7-x86_64/gcc-9/gcc-6.5.0

which is actually dead:  https://en.wikipedia.org/wiki/Scientific_Linux

Stepping back, there is a bit of a disconnect between working on old OS
releases -- sometimes not by choice as your institution, employer, $whoever
... may make you do that -- yet requiring / desiring / attempting to use
bleeding-edge newest versions of R or packages.  That can have friction as
new versions can at times require new tools. The developers try to not
require it, but sometimes it happens, and usually for a reason (as
e.g. nloptr need libnlopt which itself relies on cmake).

So if you can, please considering upgrading to a more current and supported
operating system. For issues specific with a package (as opposed to R and in
particular development on R) the repository may be a better place than this
list.

Cheers,  Dirk

--
https://dirk.eddelbuettel.com | @eddelbuettel | edd using debian.org

	[[alternative HTML version deleted]]



More information about the R-devel mailing list