Skip to content
Back to formatted view

Raw Message

Message-ID: <21148.51450.726862.110246@max.nulle.part>
Date: 2013-12-02T17:52:58Z
From: Dirk Eddelbuettel
Subject: [Rcpp-devel] Rcpp / g++ ignoring TMPDIR
In-Reply-To: <CABFfbXueRojQFjdAp7Cx+oxKwWb53c84DZhtWpq16qDJcvRqKw@mail.gmail.com>

On 2 December 2013 at 09:19, Jeroen Ooms wrote:
| On Mon, Dec 2, 2013 at 4:25 AM, Dirk Eddelbuettel <edd at debian.org> wrote:
| 
|     As I said when we hashed this out some more in private mail, I'd try
|     setting
|     TMPDIR in /etc/profile or alike. Or just try to have a working /tmp
|     directory.
|     Surely you are not the only AppArmor user with that issue?
| 
| 
| Well it's not so much an issue of AppArmor, more of using g++ with an
| process-specific TMPDIR, which I guess hasn't been tested a lot. I can easily
| work around the problem by relaxing the sandbox a bit and allowing g++ to
| create files under /tmp. I've done that on the server, and now all Rcpp
| packages build fine. However it would be more elegant if g++ would just inherit
| TMPDIR from the R process (and I am convinced it was doing that before). Anyway
| things are working again:
| 
| https://public.opencpu.org/ocpu/cran/RcppExamples
| https://public.opencpu.org/ocpu/cran/RcppArmadillo
| https://public.opencpu.org/ocpu/cran/lme4/

Thanks for circling back to close the loop.  

So no Rcpp issue, as we suspected, but possibly something in g++ et al.  That
said, availability of a sane /tmp is something most people would take for granted.

Dirk

-- 
Dirk Eddelbuettel | edd at debian.org | http://dirk.eddelbuettel.com