[R-pkg-devel] Byte-compilation problem for R-package ddalpha after update to R 3.5.0

Joris Meys Joris.Meys at ugent.be
Mon Apr 30 11:45:56 CEST 2018


Dear pavlo,

fwiw, the binary of ddalpha 1.3.2 installs just fine on Windows 10. When
building from source, I get a whole lot of warnings about uninitialized
variables, comparisons between signed and unsigned integers. The
compilation of the source code doesn't have a large impact on the memory. I
can confirm that the last step of the installation process (the byte
compiling of the actual package) requires at least 8 Gb. I have 8Gb DDR3 of
which 2.5 Gb is used by other processes. During the build process it fills
up and at one point  I have another 3 Gb in swap. That works on Windows 10
(the installation process exits successfully) but might well be the reason
why on other systems you get the errors you see.

One thing I noticed, is humongously huge code files (eg dataf.tecator.R is
a whopping 624 Kb). I would start with storing the data structures you
include as a .RDS file in a data folder (eg a file tefactor.rda containing
a data structure tefactor) , and rewrite the functions as

dataf.tefactor <- function() {return(tefactor) }

That will save in total several Mb of code files that need to be processed
by the byte compiler. Not sure this will solve the problem, but it's a low
hanging fruit ready to pick. If you're not sure what I'm talking about, you
can link me to the github repo (if you have one) and I'll take a look.

Cheers
Joris

On Mon, Apr 30, 2018 at 2:32 AM, MOZHAROVSKYI Pavlo <
pavlo.mozharovskyi at ensai.fr> wrote:

>
> After the R-update to version 3.5.0 there seems to be a problem with our
> R-package ddalpha. As it seems that the problem is connected to the
> newly introduced byte-compilation on installation, we feel unable to
> locate its root cause and ask for your help.
>
> During the byte-compilation of the package ddalpha (latest (CRAN)
> version 1.3.2), an immense amount of memory (more than 4Gb) is consumed.
> Further, byte-compilation takes at least several minutes to complete on
> Windows and MacOS, while it throws an error on the Linux system where it
> does not install at all:
>
> Error message:
> Error in system2(file.path(R.home("bin"), "R"), c(if (nzchar(arch))
> paste0("--arch=",  :
>    cannot popen ' '/usr/lib64/R/bin/R' --no-save --slave 2>&1 <
> '/tmp/RtmpU4dXOt/file3fb5b89c507'', probable reason 'Cannot allocate
> memory'
>
> We suspect a mistake in the design of our package and would appreciate
> your advice. From our side, we are ready to undertake necessary actions
> as soon as possible.
>
> Here are the existing discussions on this topic:
>
> - SO:
> https://stackoverflow.com/questions/50028566/cannot-
> allocate-memory-while-re-installing-r-package-ddalpha-under-arch-linu
> - Arch Linux: https://bbs.archlinux.org/viewtopic.php?id=236517
> - RStudio Community:
> https://community.rstudio.com/t/memory-allocation-problem-
> while-re-installing-ddalpha-package-after-upgrade-to-r-3-5-0/7663/3
>
> Thank you a lot for your help!
>
> Best regards,
> Pavlo
>
>
> --
> ------------------------------------------------------------
> Pavlo Mozharovskyi
> ENSAI - Campus de Ker-Lann
> Rue Blaise Pascal - BP 37203
> 35172 Bruz cedex
> FRANCE
>
> Tel: +33 (0)2 99 05 32 55
> Fax: +33 (0)2 99 05 32 05
> Email: pavlo.mozharovskyi at ensai.fr
>
> http://www.ensai.fr/enseignant/alias/pavlo-mozharovskyi.html
> ______________________________________________
> R-package-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-package-devel
>



-- 
Joris Meys
Statistical consultant

Department of Data Analysis and Mathematical Modelling
Ghent University
Coupure Links 653, B-9000 Gent (Belgium)
<https://maps.google.com/?q=Coupure+links+653,%C2%A0B-9000+Gent,%C2%A0Belgium&entry=gmail&source=g>

tel: +32 (0)9 264 61 79
-----------
Biowiskundedagen 2017-2018
http://www.biowiskundedagen.ugent.be/

-------------------------------
Disclaimer : http://helpdesk.ugent.be/e-maildisclaimer.php

	[[alternative HTML version deleted]]



More information about the R-package-devel mailing list