[Rd] 1.9.0 regression test on HP-UX (PR#6800)

[ripley at stats.ox.ac.uk]

All we needed was

gannet% tail reg-tests-1.Rout.fail
6           1   6
attr(,"assign")
[1] 0 1
> ##
>
>
> ## broken strptime in glibc (and code used on Windows)
> stopifnot(!is.na(strptime("2003-02-30", format="%Y-%m-%d")))
Error: !is.na(strptime("2003-02-30", format = "%Y-%m-%d")) is not TRUE
Execution halted

which makes clear that your platform has a broken implementation of 
strptime.  That's not an R bug, and no platform that anyone bothered to 
alpha or beta test failed that test ....

You may want to undefine HAVE_WORKING_STRPTIME in src/include/config.h and
build again -- it is possible that will work around this.

I trust that you will now file (in the right place) a more careful bug
report on the HP-UX error you have discovered.


On Tue, 20 Apr 2004, Paul Hatton wrote:

> Thanks. Zipped results file attached.
> 
> Regards
> 
> Date sent:      	Tue, 20 Apr 2004 13:00:24 +0100 (BST)
> From:           	Prof Brian Ripley <ripley at stats.ox.ac.uk>
> To:             	P.S.Hatton at bham.ac.uk
> Copies to:      	r-devel at stat.math.ethz.ch
> Subject:        	Re: [Rd] 1.9.0 regression test on HP-UX (PR#6800)
> 
> > The crucial information is in tests/reg-tests-1.Rout.fail, at the
> > bottom (I expect).  Please do tell us what the failure was and we may
> > be able to help.
> > 
> > Note that I have removed the bug repository from the followup: I don't
> > see any evidence of an R bug here as yet.
> > 
> > On Tue, 20 Apr 2004 P.S.Hatton at bham.ac.uk wrote:
> > 
> > > Hi,
> > > 
> > > On behalf of one of our users, I installed R 1.8.1 on HP-UX11.0 by
> > > compiling the source using the gnu compilers and all was fine with
> > > 
> > > make check
> > > 
> > > On 1.9.0, I get an error from 'make check' when running
> > > reg-tests-1.R which gives error exit code 1.  Commands used are:
> > > 
> > > ./configure --prefix=/apps/global/Gnu_R/R-1.9.0 --with-x 
> > > R_BROWSER=mozilla
> > > 
> > > (all on one line, of course)
> > > 
> > > make
> > > make check
> > > 
> > > I tried to get it to make with the HP-UX cc, aCC and f90 but
> > > couldn't get it to make (make came back with numerous unresolved
> > > symbols such as dtrsl), so I'm just using the gnu compilers. I guess
> > > that setting the loader search path would get round this, but for
> > > now at least the gnu compilers will be good enough.
> > > 
> > > Any advice appreciated.
> > > 
> > > Thanks
> > > 
> > > 
> > 
> > -- 
> > 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