This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Re: local($tied->{foo}) leaks
[perl5.git] / hints / dec_osf.sh
CommitLineData
a0d0e21e 1# hints/dec_osf.sh
8e964347 2
b971f6e4 3# * If you want to debug perl or want to send a
8e964347
AB
4# stack trace for inclusion into an bug report, call
5# Configure with the additional argument -Doptimize=-g2
6# or uncomment this assignment to "optimize":
7#
8#optimize=-g2
9#
8a019ce7
JH
10# If you want both to optimise and debug with the DEC cc
11# you must have -g3, e.g. "-O4 -g3", and (re)run Configure.
12#
b971f6e4 13# * gcc can always have both -g and optimisation on.
8a019ce7 14#
b971f6e4 15# * debugging optimised code, no matter what compiler
8a019ce7
JH
16# one is using, can be surprising and confusing because of
17# the optimisation tricks like code motion, code removal,
18# loop unrolling, and inlining. The source code and the
19# executable code simply do not agree any more while in
20# mid-execution, the optimiser only cares about the results.
21#
b971f6e4 22# * Configure will automatically add the often quoted
f2c69087
JH
23# -DDEBUGGING for you if the -g is specified.
24#
b971f6e4 25# * There is even more optimisation available in the new
f2c69087
JH
26# (GEM) DEC cc: -O5 and -fast. "man cc" will tell more about them.
27# The jury is still out whether either or neither help for Perl
28# and how much. Based on very quick testing, -fast boosts
29# raw data copy by about 5-15% (-fast brings in, among other
30# things, inlined, ahem, fast memcpy()), while on the other
31# hand searching things (index, m//, s///), seems to get slower.
32# Your mileage will vary.
8e964347 33#
b971f6e4 34# * The -std is needed because the following compiled
e60a08f8
JH
35# without the -std and linked with -lm
36#
37# #include <math.h>
38# #include <stdio.h>
39# int main(){short x=10,y=sqrt(x);printf("%d\n",y);}
40#
41# will in Digital UNIX 3.* and 4.0b print 0 -- and in Digital
42# UNIX 4.0{,a} dump core: Floating point exception in the printf(),
43# the y has become a signaling NaN.
44#
b971f6e4 45# * Compilation warnings like:
46#
47# "Undefined the ANSI standard macro ..."
48#
49# can be ignored, at least while compiling the POSIX extension
50# and especially if using the sfio (the latter is not a standard
51# part of Perl, never mind if it says little to you).
52#
8e964347 53
e60a08f8
JH
54# If using the DEC compiler we must find out the DEC compiler style:
55# the style changed between Digital UNIX (aka DEC OSF/1) 3 and
56# Digital UNIX 4. The old compiler was originally from Ultrix and
57# the MIPS company, the new compiler is originally from the VAX world
58# and it is called GEM. Many of the options we are going to use depend
59# on the compiler style.
60
a68015de
JH
61cc=${cc:-cc}
62
6c7aff00
JH
63case "`$cc -v 2>&1 | grep cc`" in
64*gcc*) isgcc=gcc ;;
65esac
66
b691c02f
JH
67# do NOT, I repeat, *NOT* take away the leading tabs
68# Configure Black Magic (TM)
b971f6e4 69 # reset
e60a08f8 70 _DEC_cc_style=
6c7aff00
JH
71case "$isgcc" in
72gcc) _gcc_version=`$cc --version 2>&1 | tr . ' '`
2cae8c0d 73 set $_gcc_version
3f8b8817 74 if test "$1" -lt 2 -o \( "$1" -eq 2 -a \( "$2" -lt 95 -o \( "$2" -eq 95 -a "$3" -lt 2 \) \) \); then
2cae8c0d
JH
75 cat >&4 <<EOF
76
61f70568
JB
77*** Your cc seems to be gcc and its version ($_gcc_version) seems to be
78*** less than 2.95.2. This is not a good idea since old versions of gcc
79*** are known to produce buggy code when compiling Perl (and no doubt for
80*** other programs, too).
3f8b8817 81***
61f70568
JB
82*** Therefore, I strongly suggest upgrading your gcc. (Why don't you use
83*** the vendor cc is also a good question. It comes with the operating
3f8b8817 84*** system and produces good code.)
2cae8c0d
JH
85
86Cannot continue, aborting.
87
88EOF
89 exit 1
90 fi
3f8b8817
JH
91 if test "$1" -eq 2 -a "$2" -eq 95 -a "$3" -le 2; then
92 cat >&4 <<EOF
93
94*** Note that as of gcc 2.95.2 (19991024) and Perl 5.6.0 (March 2000)
95*** if the said Perl is compiled with the said gcc the lib/sdbm test
61f70568
JB
96*** may dump core (meaning that the SDBM_File extension is unusable).
97*** As this core dump never happens with the vendor cc, this is most
98*** probably a lingering bug in gcc. Therefore unless you have a better
99*** gcc installation you are still better off using the vendor cc.
3f8b8817
JH
100
101Since you explicitly chose gcc, I assume that you know what are doing.
102
103EOF
104 fi
2cae8c0d 105 ;;
e60a08f8 106*) # compile something small: taint.c is fine for this.
4de7e3a2 107 ccversion=`cc -V | awk '/(Compaq|DEC) C/ {print $3}' | grep '^V'`
e60a08f8 108 # the main point is the '-v' flag of 'cc'.
52c7d5b6 109 case "`cc -v -I. -c taint.c -o taint$$.o 2>&1`" in
e60a08f8 110 */gemc_cc*) # we have the new DEC GEM CC
dfa3a3d3 111 _DEC_cc_style=new
8a019ce7 112 ;;
e60a08f8 113 *) # we have the old MIPS CC
dfa3a3d3 114 _DEC_cc_style=old
8a019ce7 115 ;;
e60a08f8
JH
116 esac
117 # cleanup
52c7d5b6 118 rm -f taint$$.o
e60a08f8
JH
119 ;;
120esac
121
b971f6e4 122# be nauseatingly ANSI
6c7aff00
JH
123case "$isgcc" in
124gcc) ccflags="$ccflags -ansi"
b971f6e4 125 ;;
126*) ccflags="$ccflags -std"
127 ;;
128esac
129
130# for gcc the Configure knows about the -fpic:
131# position-independent code for dynamic loading
132
e60a08f8
JH
133# we want optimisation
134
135case "$optimize" in
6c7aff00
JH
136'') case "$isgcc" in
137 gcc) optimize='-O3' ;;
e60a08f8 138 *) case "$_DEC_cc_style" in
6c7aff00 139 new) optimize='-O4' ;;
e60a08f8
JH
140 old) optimize='-O2 -Olimit 3200' ;;
141 esac
b971f6e4 142 ccflags="$ccflags -D_INTRINSICS"
e60a08f8 143 ;;
8e964347
AB
144 esac
145 ;;
1fc4cb55 146esac
28757baa 147
6c7aff00
JH
148# we want dynamic fp rounding mode, and we want ieee exception semantics
149case "$isgcc" in
a59bce4b
JH
150gcc) ;;
151*) case "$_DEC_cc_style" in
6c7aff00
JH
152 new) ccflags="$ccflags -fprm d -ieee" ;;
153 esac
154 ;;
155esac
156
313489a2
SB
157# Make glibpth agree with the compiler suite. Note that /shlib
158# is not here. That's on purpose. Even though that's where libc
159# really lives from V4.0 on, the linker (and /sbin/loader) won't
160# look there by default. The sharable /sbin utilities were all
161# built with "-Wl,-rpath,/shlib" to get around that. This makes
162# no attempt to figure out the additional location(s) searched by
163# gcc, since not all versions of gcc are easily coerced into
164# revealing that information.
b0362887
SB
165glibpth="/usr/shlib /usr/ccs/lib /usr/lib/cmplrs/cc"
166glibpth="$glibpth /usr/lib /usr/local/lib /var/shlib"
313489a2 167
8e964347
AB
168# dlopen() is in libc
169libswanted="`echo $libswanted | sed -e 's/ dl / /'`"
170
b971f6e4 171# libPW contains nothing useful for perl
8a019ce7
JH
172libswanted="`echo $libswanted | sed -e 's/ PW / /'`"
173
723e14d4
SB
174# libnet contains nothing useful for perl here, and doesn't work
175libswanted="`echo $libswanted | sed -e 's/ net / /'`"
176
b971f6e4 177# libbsd contains nothing used by perl that is not already in libc
8a019ce7
JH
178libswanted="`echo $libswanted | sed -e 's/ bsd / /'`"
179
b971f6e4 180# libc need not be separately listed
8a019ce7
JH
181libswanted="`echo $libswanted | sed -e 's/ c / /'`"
182
b971f6e4 183# ndbm is already in libc
184libswanted="`echo $libswanted | sed -e 's/ ndbm / /'`"
8a019ce7
JH
185
186# the basic lddlflags used always
187lddlflags='-shared -expect_unresolved "*"'
188
b971f6e4 189# Fancy compiler suites use optimising linker as well as compiler.
190# <spider@Orb.Nashua.NH.US>
b8f0c030 191case "`uname -r`" in
b971f6e4 192*[123].*) # old loader
193 lddlflags="$lddlflags -O3"
194 ;;
a07564da
KS
195*) if $test "X$optimize" = "X$undef"; then
196 lddlflags="$lddlflags -msym"
197 else
baa8820a 198 case "`/usr/sbin/sizer -v`" in
e3159d07
JH
199 *4.0D*)
200 # QAR 56761: -O4 + .so may produce broken code,
201 # fixed in 4.0E or better.
202 ;;
203 *)
204 lddlflags="$lddlflags $optimize"
205 ;;
206 esac
207 # -msym: If using a sufficiently recent /sbin/loader,
208 # keep the module symbols with the modules.
baa8820a 209 lddlflags="$lddlflags -msym -std"
a07564da 210 fi
b971f6e4 211 ;;
212esac
213# Yes, the above loses if gcc does not use the system linker.
214# If that happens, let me know about it. <jhi@iki.fi>
215
8a019ce7 216
b971f6e4 217# If debugging or (old systems and doing shared)
218# then do not strip the lib, otherwise, strip.
219# As noted above the -DDEBUGGING is added automagically by Configure if -g.
8a019ce7
JH
220case "$optimize" in
221 *-g*) ;; # left intentionally blank
b8f0c030 222*) case "`uname -r`" in
b971f6e4 223 *[123].*)
224 case "$useshrplib" in
225 false|undef|'') lddlflags="$lddlflags -s" ;;
226 esac
227 ;;
8a019ce7 228 *) lddlflags="$lddlflags -s"
b971f6e4 229 ;;
230 esac
231 ;;
8a019ce7 232esac
8e964347
AB
233
234#
313489a2
SB
235# Make embedding in things like INN and Apache more memory friendly.
236# Keep it overridable on the Configure command line, though, so that
237# "-Uuseshrplib" prevents this default.
238#
239
2bf2710f
GS
240case "$_DEC_cc_style.$useshrplib" in
241 new.) useshrplib="$define" ;;
242esac
313489a2 243
85ab1d1d
JH
244# The EFF_ONLY_OK from <sys/access.h> is present but dysfunctional for
245# [RWX]_OK as of Digital UNIX 4.0[A-D]?. If and when this gets fixed,
246# please adjust this appropriately. See also pp_sys.c just before the
247# emulate_eaccess().
5ff3f7a4 248
baa8820a
JH
249# Fixed in V5.0A.
250case "`/usr/sbin/sizer -v`" in
f60a5d46 251*5.0[A-Z]*|*5.[1-9]*|*[6-9].[0-9]*)
baa8820a
JH
252 : ok
253 ;;
254*)
255# V5.0 or previous
5ff3f7a4 256pp_sys_cflags='ccflags="$ccflags -DNO_EFF_ONLY_OK"'
baa8820a
JH
257 ;;
258esac
5ff3f7a4 259
6b8eaf93
JH
260# The off_t is already 8 bytes, so we do have largefileness.
261
f60a5d46 262cat > UU/usethreads.cbu <<'EOCBU'
104d25b7
JH
263# This script UU/usethreads.cbu will get 'called-back' by Configure
264# after it has prompted the user for whether to use threads.
104d25b7
JH
265case "$usethreads" in
266$define|true|[yY]*)
75d72f2c 267 # Threads interfaces changed with V4.0.
6c7aff00
JH
268 case "$isgcc" in
269 gcc) ccflags="-D_REENTRANT $ccflags" ;;
75d72f2c
JH
270 *) case "`uname -r`" in
271 *[123].*) ccflags="-threads $ccflags" ;;
272 *) ccflags="-pthread $ccflags" ;;
273 esac
104d25b7 274 ;;
75d72f2c
JH
275 esac
276 case "`uname -r`" in
277 *[123].*) libswanted="$libswanted pthreads mach exc c_r" ;;
278 *) libswanted="$libswanted pthread exc" ;;
279 esac
e883c329 280
baa8820a
JH
281 case "$usemymalloc" in
282 '')
283 usemymalloc='n'
284 ;;
285 esac
a48ec845
JH
286 # These symbols are renamed in <time.h> so
287 # that the Configure hasproto doesn't see them.
288 d_asctime_r_proto="$define"
289 d_ctime_r_proto="$define"
290 d_gmtime_r_proto="$define"
291 d_localtime_r_proto="$define"
104d25b7
JH
292 ;;
293esac
294EOCBU
295
f60a5d46
JH
296cat > UU/uselongdouble.cbu <<'EOCBU'
297# This script UU/uselongdouble.cbu will get 'called-back' by Configure
298# after it has prompted the user for whether to use long doubles.
299case "$uselongdouble" in
bef5f079
JH
300$define|true|[yY]*)
301 case "`/usr/sbin/sizer -v`" in
302 *[1-4].0*) cat >&4 <<EOF
303
304***
305*** Sorry, you cannot use long doubles in pre-V5.0 releases of Tru64.
306***
307
308Cannot continue, aborting.
309
310EOF
311 exit 1
312 ;;
fa17d112
SB
313 *)
314 # Test whether libc's been fixed yet.
315 cat >try.c <<\TRY
316#include <stdio.h>
317int main(int argc, char **argv)
318{
319 unsigned long uvmax = ~0UL;
320 long double ld = uvmax + 0.0L;
321 char buf1[30], buf2[30];
322
323 (void) sprintf(buf1, "%lu", uvmax);
324 (void) sprintf(buf2, "%.0Lf", ld);
325 return strcmp(buf1, buf2) != 0;
326}
327TRY
328 # Don't bother trying to work with Configure's idea of
329 # cc and the various flags. This might not work as-is
330 # with gcc -- but we're testing libc, not the compiler.
331 if cc -o try -std try.c && ./try
332 then
333 : ok
334 else
335 cat <<\UGLY >&4
336!
337Warning! Your libc has not yet been patched so that its "%Lf" format for
338printing long doubles shows all the significant digits. You will get errors
339in the t/op/numconvert test because of this. (The data is still good
340internally, and the "%e" format of printf() or sprintf() in perl will still
341produce valid results.) See README.tru64 for additional details.
342
343Continuing anyway.
344!
345UGLY
346 fi
347 $rm -f try try.c
bef5f079 348 esac
bef5f079 349 ;;
f60a5d46
JH
350esac
351EOCBU
352
1cade9fc 353case "`/usr/sbin/sizer -v`" in
bef5f079 354*[1-4].0*) d_modfl=undef ;; # must wait till 5.0
1cade9fc
JH
355esac
356
c93bd003
JH
357case "$loclibpth" in
358'') ;;
359*)
360 needusrshlib=''
95fecce1
JH
361 case "$loclibpth" in
362 '') ;;
363 *) for p in $loclibpth
364 do
365 if test -n "`ls $p/libdb.so* 2>/dev/null`"; then
366 needusrshlib=yes
367 fi
368 if test -d $p; then
369 echo "Appending $p to LD_LIBRARY_PATH." >& 4
370 case "$LD_LIBRARY_PATH" in
371 '') LD_LIBRARY_PATH=$p ;;
372 *) LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$p ;;
373 esac
374 fi
375 done
376 echo "LD_LIBRARY_PATH is now $LD_LIBRARY_PATH." >& 4
377 ;;
a2e4e9f1 378 esac
c93bd003
JH
379 # This is evil but I can't think of a nice workaround:
380 # the /usr/shlib/libdb.so needs to be seen first,
381 # or running Configure will fail.
382 if test -n "$needusrshlib"; then
383 echo "Prepending /usr/shlib to loclibpth." >& 4
384 loclibpth="/usr/shlib $loclibpth"
385 echo "loclibpth is now $loclibpth." >& 4
386 fi
387 ;;
388esac
389
313489a2 390#
e60a08f8
JH
391# Unset temporary variables no more needed.
392#
393
394unset _DEC_cc_style
395
396#
8e964347
AB
397# History:
398#
85ab1d1d
JH
399# perl5.005_51:
400#
401# September-1998 Jarkko Hietaniemi <jhi@iki.fi>
402#
403# * Added the -DNO_EFF_ONLY_OK flag ('use filetest;' support).
404#
313489a2
SB
405# perl5.004_57:
406#
407# 19-Dec-1997 Spider Boardman <spider@Orb.Nashua.NH.US>
408#
85ab1d1d 409# * Newer Digital UNIX compilers enforce signaling for NaN without
313489a2
SB
410# -ieee. Added -fprm d at the same time since it's friendlier for
411# embedding.
412#
413# * Fixed the library search path to match cc, ld, and /sbin/loader.
414#
415# * Default to building -Duseshrplib on newer systems. -Uuseshrplib
416# still overrides.
417#
418# * Fix -pthread additions for useshrplib. ld has no -pthread option.
419#
420#
723e14d4
SB
421# perl5.004_04:
422#
423# 19-Sep-1997 Spider Boardman <spider@Orb.Nashua.NH.US>
424#
425# * libnet on Digital UNIX is for JAVA, not for sockets.
426#
427#
b971f6e4 428# perl5.003_28:
429#
430# 22-Feb-1997 Jarkko Hietaniemi <jhi@iki.fi>
431#
432# * Restructuring Spider's suggestions.
433#
434# * Older Digital UNIXes cannot handle -Olimit ... for $lddlflags.
435#
436# * ld -s cannot be used in older Digital UNIXes when doing shared.
437#
438#
439# 21-Feb-1997 Spider Boardman <spider@Orb.Nashua.NH.US>
440#
441# * -hidden removed.
442#
443# * -DSTANDARD_C removed.
444#
445# * -D_INTRINSICS added. (that -fast does not seem to buy much confirmed)
446#
447# * odbm not in libc, only ndbm. Therefore dbm back to $libswanted.
448#
449# * -msym for the newer runtime loaders.
450#
451# * $optimize also in $lddflags.
452#
453#
e60a08f8
JH
454# perl5.003_27:
455#
456# 18-Feb-1997 Jarkko Hietaniemi <jhi@iki.fi>
457#
458# * unset _DEC_cc_style and more commentary on -std.
459#
460#
dfa3a3d3
CS
461# perl5.003_26:
462#
463# 15-Feb-1997 Jarkko Hietaniemi <jhi@iki.fi>
464#
e60a08f8 465# * -std and -ansi.
dfa3a3d3
CS
466#
467#
f2c69087
JH
468# perl5.003_24:
469#
470# 30-Jan-1997 Jarkko Hietaniemi <jhi@iki.fi>
471#
472# * Fixing the note on -DDEBUGGING.
473#
474# * Note on -O5 -fast.
475#
476#
8a019ce7
JH
477# perl5.003_23:
478#
479# 26-Jan-1997 Jarkko Hietaniemi <jhi@iki.fi>
480#
481# * Notes on how to do both optimisation and debugging.
482#
483#
484# 25-Jan-1997 Jarkko Hietaniemi <jhi@iki.fi>
485#
486# * Remove unneeded libraries from $libswanted: PW, bsd, c, dbm
487#
488# * Restructure the $lddlflags build.
489#
490# * $optimize based on which compiler we have.
491#
492#
8e964347
AB
493# perl5.003_22:
494#
495# 23-Jan-1997 Achim Bohnet <ach@rosat.mpe-garching.mpg.de>
496#
497# * Added comments 'how to create a debugging version of perl'
498#
499# * Fixed logic of this script to prevent stripping of shared
500# objects by the loader (see ld man page for -s) is debugging
501# is set via the -g switch.
502#
503#
504# 21-Jan-1997 Achim Bohnet <ach@rosat.mpe-garching.mpg.de>
505#
506# * now 'dl' is always removed from libswanted. Not only if
507# optimize is an empty string.
508#
509#
510# 17-Jan-1997 Achim Bohnet <ach@rosat.mpe-garching.mpg.de>
511#
512# * Removed 'dl' from libswanted: When the FreePort binary
513# translator for Sun binaries is installed Configure concludes
514# that it should use libdl.x.yz.fpx.so :-(
515# Because the dlopen, dlclose,... calls are in the
516# C library it not necessary at all to check for the
517# dl library. Therefore dl is removed from libswanted.
518#
519#
520# 1-Jan-1997 Achim Bohnet <ach@rosat.mpe-garching.mpg.de>
521#
522# * Set -Olimit to 3200 because perl_yylex.c got too big
523# for the optimizer.
524#