On Wed, 19 Feb 2003, Zhang, Sonic wrote:
> Hi,
>
> I tried the pcp test suite v1.1 on United Linux and got some
> failures in case 252, 261, 280.
252 your gawk has become (absurdly, IMHO) picky .. provided
\. is interpreted as . then I'm not interested in the warning
... is there some way to make gawk behave sensibly?
if not, you'll need this patch
rattle 60% p_rdiff -u 252
--- /usr/tmp/TmpDir.14250867-0/qa/252_1.8 Wed Feb 19 14:12:47 2003
+++ qa/252 Wed Feb 19 13:59:04 2003
@@ -123,7 +123,7 @@
/usr/bin/time $* 2>&1 >/dev/null | \
if [ "$PCP_PLATFORM" = "linux" ] ; then
tr ' ' "\n" | $PCP_AWK_PROG '/elapsed$/ { sub("elapsed", "", $1);
- split ($1,tt,"[:\.]");
+ split ($1,tt,"[:.]");
print (tt[1]*60)+tt[2];}'
else
$PCP_AWK_PROG '/^real/ {print $2}' | sed -e 's/\..*//'
261 I can reporoduce this failure case on Red Hat 7.3 locally,
so I will investigate
280 your ls(1) is producing dates like "1997-01-02 03:04" instead
of "Jan 2 1997" ... can you make ls(1) behave the way it should
(again IMHO)
TIme to climb on my soap box ...
I'm afraid I do not have time to pander to a whole raft of QA
failures that are not PCP QA "failures" but in my opinion United
Linux (or indeed any other brand of Linux/Unix) "failures" because
the distributors, or the contributors are making arbitrary
decisions to change the way long standing Unix utilities behave
... if you have these problems, then the onus is on those with a
vested interest in these operating system versions to send me
patches to fix the PCP QA suite ... if these patches work for IRIX
and the Linux distribution I care about, I'll take them, otherwise
I suggest you get your operating system fixed.
For the cases where there is some suggestion of a genuine PCP
problem, I will continue to investigate, as in 261 above.
|