[Rd] readBin should check that its endian argument is a legal value
jenn||er@@@|yon @end|ng |rom gm@||@com
Wed Dec 18 17:12:03 CET 2019
On Wed, Dec 18, 2019 at 4:12 AM Tomas Kalibera <tomas.kalibera using gmail.com>
> Thank you for reporting this problem, R-devel now has a check in readBin
> and writeBin.
> I've identified two CRAN packages with an incorrect value for "endian"
> and reported to maintainers, unfortunately in their case the intention
> was to specify "little".
> On 11/18/19 11:22 PM, Jennifer Lyon wrote:
> > I think it would be helpful if readBin checked that its endian argument
> > a legal value.
> > Why? I was reviewing some of our code and noticed that the author had
> > readBin(..., endian="network") and never having heard of "network", I
> > looked at the man page for readBin, and it hadn't heard of "network"
> > either. Not good.
> > I then looked at the R code for readBin, which has this line:
> > swap <- endian != .Platform$endian
> > and swap is passed into the .Internal(readBin). Further use of Google
> > revealed that "network" is a known endian in the universe, and our code
> > working by essentially a lucky chance that the data was "big" and our
> > current machines are "little". Really not good. I don't know enough about
> > endian stuff to know if it makes sense that "network" should be one of
> > choices for endian for readBin (which from the documentation currently
> > "big", "little" or "swap"), but in my opinion R should have failed with
> > choice of "network" in our code in the current version of R. I did
> > eventually find an aside in the R Data Import/Export document that
> > "network" is "big" so I will patch our code to be legal. Of course some
> > code may depend on this undocumented behavior, but I would guess that in
> > the vast majority of cases an illegal value really is a mistake.
> > Jen
> >> sessionInfo()
> > R version 3.6.1 (2019-07-05)
> > Platform: x86_64-pc-linux-gnu (64-bit)
> > Running under: Ubuntu 18.04.3 LTS
> > Matrix products: default
> > BLAS: /home/mbr/r-project/R-3.6.1/lib/libRblas.so
> > LAPACK: /home/mbr/r-project/R-3.6.1/lib/libRlapack.so
> > locale:
> >  LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C
> >  LC_TIME=en_US.UTF-8 LC_COLLATE=en_US.UTF-8
> >  LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8
> >  LC_PAPER=en_US.UTF-8 LC_NAME=C
> >  LC_ADDRESS=C LC_TELEPHONE=C
> >  LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C
> > attached base packages:
> >  stats graphics grDevices utils datasets methods base
> > loaded via a namespace (and not attached):
> >  compiler_3.6.1
> > [[alternative HTML version deleted]]
> > ______________________________________________
> > R-devel using r-project.org mailing list
> > https://stat.ethz.ch/mailman/listinfo/r-devel
[[alternative HTML version deleted]]
More information about the R-devel