[Rd] R-patched Make Check Fails on reg-tests-1.R on linux and OS X

[Prof Brian Ripley]

Stefano Iacus has been able to reproduce this on MacOS X (it was a 
segfault in a later example), and we have just now committed a workaround.
I think it probably was an OS-specific quirk in vsnprintf.

On Tue, 8 Feb 2005, Prof Brian Ripley wrote:

> Since no one else has this problem, I suggest you check the integrity of your 
> checkout, or, better, use an R-patched tarball that can easily be verified. 
> This looks very like a mismatched build and test: that is your R build has 
> not be updated to include the patch which is being tested.
> A completely clean build from a tarball will ensure that is not the case.
>
> Reporting problems using unreleased compilers (gcc 3.4.4 is not released) 
> isn't going to win you a lot of sympathy: they have been responsible for a 
> large number of (mis-directed) problem reports.  If you didn't have the 
> problem on two machines I would be suggesting using released versions of the 
> tools.
>
> FYI, R is tested on released compilers on i686 Linux several times a day, and 
> at least weekly on other common platforms.  We would know soon enough if 
> there was an R problem in 'make all check' on those platforms.
>
>
> On Mon, 7 Feb 2005, Jake Bowers wrote:
>
>> Dear Developers,
>> 
>> I've been playing around with compiling R on my Debian Linux machine (dual
>> Athlon 1.4ghz) and my OS X machine (dual G5). I'm emailing now because
>> reg-tests-1.R fails during make check on my debian machine using gcc-3.4,
>> and on my OS X machine using gcc-3.3. I am using r-patched updated via svn
>> today (Updated to revision 33075.)
>> 
>> Here are some details:
>> 
>> **Using gcc-3.4 on debian:
>> gcc-3.4 (GCC) 3.4.4 20041218 (prerelease) (Debian 3.4.3-6)
>> 
>> wes:/home/temp/R/r-patched/tests# tail reg-tests-1.Rout.fail
>>> 
>>> ## automatic row.names can be number-like, MM, 2004-11-26
>>> d0 <- data.frame(x=1:3, y=pi*2:0)
>>> row.names(d0)[3] <- c("01.00")
>>> write.table(d0, (tf <- tempfile()))
>>> d <- read.table(tf)
>>> ## gave error ("duplicate row.names") in 2.0.1
>>> stopifnot(all.equal(d,d0))
>>> unlink(tf)
>> 
>> **Using gcc-3.3 on debian works fine (passes all make check).
>> gcc (GCC) 3.3.5 (Debian 1:3.3.5-5)
>> 
>> **Using gcc-3.3 on OS X.
>> gcc (GCC) 3.3 20030304 (Apple Computer, Inc. build 1671)
>> 
>> More info on my OS X build:
>> ./configure --with-blas='-framework vecLib' --with-lapack --with-aqua
>> --with-x --with-tcl-config=/Library/Frameworks/Tcl.framework/tclConfig.sh
>> --with-tk-config=/Library/Frameworks/Tk.framework/tkConfig.sh
>> --enable-R-shlib TCLTK_LIBS='-framework Tcl -framework Tk'
>> TCLTK_CPPFLAGS='-I/Library/Frameworks/Tcl.Framework/Headers
>> -I/Library/Frameworks/Tk.Framework/Headers' --with-recommended
>> 
>> g77 is version 3.4 downloaded from hpc.sf.net.
>> GNU Fortran (GCC) 3.4.2
>> 
>> echo $PATH
>> /usr/local/bin:/bin:/sbin:/usr/bin:/usr/sbin:/usr/X11R6/bin:/sw/bin:/sw/sbin:/usr/local/pvm3/lib:/usr/local/pvm3/bin/DARWIN
>> 
>> and, to prevent it from using stuff in the fink directory:
>> 
>> CPPFLAGS='-I/usr/local/include'
>> 
>> Here is the output where make check fails:
>> 
>> running regression tests
>> running code in 'reg-tests-1.R' ...make[3]: *** [reg-tests-1.Rout] Error 1
>> make[2]: *** [test-Reg] Error 2
>> make[1]: *** [test-all-basics] Error 1
>> make: *** [check-all] Error 2
>> 
>> sphere:~/TEMP/R/r-patched/tests jwbowers$ tail reg-tests-1.Rout.fail
>>> 
>>> 
>>> ## automatic row.names can be number-like, MM, 2004-11-26
>>> d0 <- data.frame(x=1:3, y=pi*2:0)
>>> row.names(d0)[3] <- c("01.00")
>>> write.table(d0, (tf <- tempfile()))
>>> d <- read.table(tf)
>>> ## gave error ("duplicate row.names") in 2.0.1
>>> stopifnot(all.equal(d,d0))
>>> unlink(tf)
>> 
>> Should I be very concerned about this? I tend to mostly use my OS X
>> machine since the Linux box is about 4 years old.
>> 
>> I hope this information is helpful --- I'm sorry if this is something
>> obvious! (I found some posts from last summer about problems with gcc-3.4,
>> which might explain the problems with gcc-3.4 on linux, but I didn't find
>> anything obvious about gcc-3.3 on the Mac).
>> 
>> Thanks so much for all of your work!!
>> 
>> Best,
>> 
>> Jake
>> 
>> Jake Bowers
>> Assistant Professor
>> Dept of Political Science
>> University of Michigan
>> http://www.umich.edu/~jwbowers
>> 
>> ______________________________________________
>> R-devel at stat.math.ethz.ch mailing list
>> https://stat.ethz.ch/mailman/listinfo/r-devel
>> 
>> 
>
> -- 
> Brian D. Ripley,                  ripley at stats.ox.ac.uk
> Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
> University of Oxford,             Tel:  +44 1865 272861 (self)
> 1 South Parks Road,                     +44 1865 272866 (PA)
> Oxford OX1 3TG, UK                Fax:  +44 1865 272595
>

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595