On Fri, Jan  8, 2016 at 08:59:13 -0600, Dirk Eddelbuettel wrote:

> 
> Hi Seb,
> 
> On 8 January 2016 at 14:54, Sébastien Villemot wrote:
> | Package: r-base-core
> | Version: 3.2.3-1
> | Severity: grave
> 
> That will deprive a couple (ten or hundred) thousand users of R because ...
> 
> | Control: affects -1 r-cran-ncdf4 r-cran-afex r-cran-coin r-cran-scales 
> r-cran-ape r-cran-rms r-cran-timeseries r-cran-dosefinding r-cran-maps 
> r-cran-statmod
> | 
> | -----BEGIN PGP SIGNED MESSAGE-----
> | Hash: SHA256
> | 
> | Hi Dirk,
> | 
> | R crashes at startup with an "Illegal instruction" (tested on minkus, the 
> mips
> | porterbox):
> | 
> |   (sid_mips-dchroot)sebastien@minkus:~$ R
> |   Illegal instruction
> | 
> | Here is a backtrace:
> | 
> | #0  machar (ibeta=0x77fc3bd0 <R_AccuracyInfo>, it=0x77fc3bd4 
> <R_AccuracyInfo+4>, irnd=0x77fc3bd8 <R_AccuracyInfo+8>, ngrd=0x77fc3bdc 
> <R_AccuracyInfo+12>, machep=0x77fc3be0 <R_AccuracyInfo+16>,
> |     negep=0x77fc3be4 <R_AccuracyInfo+20>, iexp=0x77fc3be8 
> <R_AccuracyInfo+24>, minexp=0x77fc3bec <R_AccuracyInfo+28>, maxexp=0x77fc3bf0 
> <R_AccuracyInfo+32>, eps=0x77fc3bf8 <R_AccuracyInfo+40>,
> |     epsneg=0x77fc3c00 <R_AccuracyInfo+48>, xmin=0x77fc3c08 
> <R_AccuracyInfo+56>, xmax=0x77fc3c10 <R_AccuracyInfo+64>) at platform.c:3199
> | #1  0x77d1a570 in Init_R_Machine (rho=0x43054c) at platform.c:61
> | #2  0x77d1b48c in Init_R_Variables (rho=0x43054c) at platform.c:198
> | #3  0x77cf0198 in setup_Rmainloop () at main.c:840
> | #4  0x77cf0dfc in Rf_mainloop () at main.c:1012
> | #5  0x00400a00 in main (ac=1, av=0x7fff6574) at Rmain.c:29
> | #6  0x779b9f58 in __libc_start_main () from /lib/mips-linux-gnu/libc.so.6
> | #7  0x00400870 in _ftext ()
> | 
> | Incidentally, all CRAN packages FTBFS on mips (see above for the list of
> | currently affected packages).
> 
> ... neither I not R Core have mips access, as this may become
> release-critical and block R.  :-/
> 
This is untrue.  You have access to debian's mips porterbox.

Cheers,
Julien
-- 
Julien Cristau          <julien.cris...@logilab.fr>
Logilab                 http://www.logilab.fr/
Informatique scientifique & gestion de connaissances

Reply via email to