configure bug with IRIX 6.5 (Makedep) (PR#288)

caves@yorvic.york.ac.uk caves@yorvic.york.ac.uk
Tue, 28 Sep 1999 10:30:55 +0200 (MET DST)


Full_Name: Leo Caves
Version: 0.65.0
OS: SGI/IRIX 6.5
Submission from: (NULL) (144.32.72.76)


using default configure under SGI/IRIX 6.5.
(I switched to gnumake, after initial problems, this might not be necesssary).

the make rule for creating dependencies is wrong for the SGI C compiler.
The rule assumes that cc -M file.c -o file.d works (it doesn't).
One needs to explicitly capture stdout to grab the dependencies:
i.e. cc -M file.c > file.d

simple as that.

when I run tests, there are 2 errors reported

1/ in print-tests.R
(printing *very* small numbers with the statement:)
> cat("dig|  formatC(signif(txn, d=dig)\n")
dig|  formatC(signif(txn, d=dig)
> for(dig in 1:14)#15: Linux 2.1.120, libc-2.0.7 differs from Debian 2.0
+     cat(formatC(dig,w=2), formatC(signif(txn, d=dig), dig=dig+2,
wid=-20),"\n")

Where the blessed version has 0 (zero), the SGI version shows 2.e-308

This looks pretty benign and is flagged in the test script as dependent on
specific libraries etc under linux.

2/ in simple-tests.R
(comparing numbers at limit of word precision)
abs(1- .Machine$double.xmax * 10^(-.Machine$double.max.exp*log10(2)))/Meps <
1e3

This statement returns F instead of T.

This may have effects in pathological cases, but doesn't seem to be a
show-stopper.

So, it looks like a kosher build under SGI/IRIX 6.5.

*Many thanks* for a great system.
Leo
(vanilla AT&T S user since 1991)



-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._