[ksh93-integration-discuss] "iszero()" gone in ast-ksh/2006-11-22 ?

[Roland Mainz]

Hi!

----

Is it possible that the builtin fuction "iszero" somehow disappeared in
ast-ksh/2006-11-22 ?

$ ./arch/sol11.i386/bin/ksh -o gmacs     
$ (float x=0 ; print $(( iszero(x) )) )
/home/test001/tmp/build32/arch/sol11.i386/bin/ksh: iszero(x) : unknown
function

A quick look at the build log shows this:
-- snip --
iffe: test: is isnan a library function ... yes
iffe: test: is isnanl a library function ... yes
iffe: test: is isnormal a library function ... yes
iffe: test: is isnormall a library function ... no
iffe: test: is issubnormal a library function ... no
iffe: test: is issubnormall a library function ... no
iffe: test: is isunordered a library function ... yes
iffe: test: is isunorderedl a library function ... no
iffe: test: is iszero a library function ... no
iffe: test: is iszerol a library function ... no
iffe: test: is lgamma a library function ... yes
iffe: test: is lgammal a library function ... yes
-- snip --
and
-- snip --
iffe: test: is sys/types.h a header ... yes
iffe: test: is math.h a header ... yes
iffe: test: is ieeefp.h a header ... yes
iffe: test: is finitel a macro ... no
iffe: test: is fpclassifyl a macro ... no
iffe: test: is isfinitel a macro ... no
iffe: test: is isgreaterl a macro ... no
iffe: test: is isgreaterequall a macro ... no
iffe: test: is isinfl a macro ... no
iffe: test: is islessl a macro ... no
iffe: test: is islessequall a macro ... no
iffe: test: is islessgreaterl a macro ... no
iffe: test: is isnormall a macro ... no
iffe: test: is issubnormall a macro ... no
iffe: test: is issubnormal a macro ... no
iffe: test: is isunorderedl a macro ... no
iffe: test: is iszerol a macro ... no
iffe: test: is iszero a macro ... no
iffe: test: is signbitl a macro ... no
iffe: test: capture output of
/home/test001/tmp/build32/src/cmd/ksh93/features/math.sh ... yes
-- snip --

Steps to reproduce (for i386/AMD&4):
$ wget
http://svn.genunix.org/repos/on/branches/ksh93/gisburn/scripts/buildksh93.ksh
$ wget --http-user="I accept www.opensource.org/licenses/cpl"
--http-passwd="."
'http://www.research.att.com/~gsf/download/alpha/INIT.2006-11-22.tgz'
$ wget --http-user="I accept www.opensource.org/licenses/cpl"
--http-passwd="."
'http://www.research.att.com/~gsf/download/alpha/ast-ksh.2006-11-22.tgz'
$ mkdir build32 ; cd build32
$ gunzip -c ../ast-ksh.2006-11-22.tgz | tar -xf -
$ gunzip -c ../INIT.2006-11-22.tgz | tar -xf -
$ time nice ksh ../buildksh93.ksh build.solaris.i386.32bit.suncc 2>&1 |
tee -a buildlog.log
$ ./arch/sol11.i386/bin/ksh -o gmacs     
$ (float x=0 ; print $(( iszero(x) )) )
./arch/sol11.i386/bin/ksh iszero(x) : unknown function

This looks a little bit weired... "isnormal" is there, "issubnormal"
isn't... and I don't see a pattern why one is there and the other
isn't... the "buildksh.ksh" script enforces C99 mode via
"/opt/SUNWspro/bin/cc -xc99=%all -D_XOPEN_SOURCE=600 -D__EXTENSIONS__=1
..." and should make all these functions accessible to the code... ;-/


Slightly offtopic:
Trying to use "isunordered" results in a core dump:
-- snip --
$ (float x=-1 ; print $(( isunordered(x, 1) )) )
Memory fault(coredump)
$ pstack core
core 'core' of 18197:  
/home/test001/tmp/build32/arch/sol11.i386/bin/ksh -o gmacs
 75732700 ???????? (8203c78)
 08101767 strval   (8203c11, 8046c48, 80f6bd0, 1) + 87
 080f78f4 sh_strnum (8203c11, 0, 1) + b4
 080f79ae sh_arith (8203c11) + 2e
 080af751 comsubst (81fd978, 1) + 1b1
 080acfdf varsub   (81fd978) + 2bf
 080ab81f copyto   (81fd978, 0, 0) + 8df
 080aa0f0 sh_macexpand (8203bcc, 80471a4, 0) + 2d0
 080f9dcd arg_expand (8203bcc, 80471a4, 0) + 2ed
 080f98fe sh_argbuild (80473b0, 8203bac, 0) + 11e
 080cdee9 sh_exec  (8203bac, 4) + 259
 080d0232 sh_exec  (8203bac, 4) + 25a2
 080c9dc4 sh_subshell (8203bf0, 4, 0) + 534
 080cfffe sh_exec  (8203b04, 4) + 236e
 0808b741 exfile   (81ec5ac, 81eea5c, 0) + b81
 0808ab89 sh_main  (3, 8047b94, 0) + b39
 0808977a main     (3, 8047b94, 8047ba4) + 3a
 0808956a _start   (3, 8047c78, 8047caa, 8047c78, 0, 81fd938) + 7a
-- snip --
... and dbx is defunct in B51 thanks to the brandz zone symbol issue...
fun... ;-(

----

Bye,
Roland

P.S.: Chris: Is there any ETA when a fix for the dbx problem becomes
available ?

-- 
  __ .  . __
 (o.\ \/ /.o) roland.mainz at nrubsig.org
  \__\/\/__/  MPEG specialist, C&&JAVA&&Sun&&Unix programmer
  /O /==\ O\  TEL +49 641 7950090
 (;O/ \/ \O;)