[R] shared object location

[Murat Tasan]

as a motivating example, consider the case where multiple/many R
packages are being developed that all use the same shared object
functions.

each time (during development) that the shared object file changes we
could go alter every single R package directory, or simply have the R
package utilize the most recent single copy of the shared object's
functions.
this is essentially what happens already with most programs and the
common libraries in /usr/lib, /usr/local/lib, etc.

but during development of the library, it's preferable to keep this
library out of those system-level trees, and instead somewhere more
local (e.g. in the developer's home directory, or a location set aside
by a team of developers, e.g. /dev/lib).

then i would like for each of the R packages to do something like
'library.dynam("testlib1")', looking along the LD_LIBRARY_PATH for the
R session (which will include /dev/lib), for the library /dev/lib/
testlib1.so.

but it appears library.dynam can ONLY search inside package
contents... is this correct?
i can keep telling every package writer to include 'dyn.load("/dev/lib/
testlib1.so")' somewhere in their package, but this is fragile (as
dyn.load requires an exact path to be specified, instead of using
LD_LIBRARY_PATH).

so it appears i either have fragility, or an error-prone process of
updating the .so files in each R package separately each time the
library changes (which is frequently).
i have to imagine there is some method in R that i just cannot
identify that will allow for linking/loading of shared objects along
the LD_LIBRARY_PATH.


On Jan 30, 5:52 pm, Murat Tasan <mmu... at gmail.com> wrote:
> The primary reason is for development.
> I do install via the traditional R CMD INSTALL (or variant) method,
> but I want to keep the C code external from the R package.
> In particular, I want to be able to modify the C code (and thus the
> compiled .so library functions) without having to constantly re-
> install the R package, or constantly copy the shared object library to
> the /libs/ subdirectory in the R package directory.
>
> The crux of the matter is that this seems do-able: it basically works
> using dyn.load, just not easily with library.dynam, which is what
> confuses me.
> The vast majority of other UNIX-ish applications will somewhere allow
> me to specify a non-standard library location, if needed
> In many cases, this is done (in an ad-hoc, and thus unstable manner)
> using LD_LIBRARY_PATH, but this doesn't seem to be working for me in
> this instance.
>
> On Jan 29, 4:53 pm, Seth Falcon <s... at userprimary.net> wrote:
>
>
>
>
>
> > Hi,
>
> > On 1/29/10 12:58 PM,MuratTasan wrote:
>
> > > problem is, i haven't been able to do this yet.
>
> > > my workaround is to constantly swap into myRPackage/libs/ directory a
> > > version of the shared object library called myRPackage.so, and load it
> > > via the useDynLib(...) directive in the NAMESPACE file of the
> > > package.  but this is cumbersome and doesn't allow other people to
> > > have a single version of the R package to use as part of a team effort
> > > to debug/test both the package's R code and my C code for the library
> > > functions.
>
> > > hopefully this makes a bit more sense, and if anyone has any tips on
> > > how R actually loads shared objects it would be greatly appreciated.
>
> > I'm afraid I don't have a solution for you.  Reading through your post,
> > I'm confused about why you want to have your .so file outside of the
> > package?
>
> > Why not follow the usual scheme and R CMD INSTALL yourPkg with the
> > version of R code and C code you want to use/test.  This would best
> > allow other people to have a single version of the R package, no?
>
> > + seth
>
> > --
> > Seth Falcon | @sfalcon |http://userprimary.net/user
>
> > ______________________________________________
> > R-h... at r-project.org mailing listhttps://stat.ethz.ch/mailman/listinfo/r-help
> > PLEASE do read the posting guidehttp://www.R-project.org/posting-guide.html
> > and provide commented, minimal, self-contained, reproducible code.
>
> ______________________________________________
> R-h... at r-project.org mailing listhttps://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guidehttp://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.