From: Perl 5 Porters Date: Mon, 24 Jun 1996 03:07:53 +0000 (+0000) Subject: perl 5.003: hints/linux.sh X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=commitdiff_plain;h=ef0c99469b039696f8730cf24a8f0c48662abffd;p=p5sagit%2Fp5-mst-13.2.git perl 5.003: hints/linux.sh Allow Configure to determine malloctype Don't mess with $nm_so_opt Make sure messages get to the right places Disable NDBM, since some systems have problems with the libraries Don't use gcvt() to print floats; it's buggy in some versions of libc --- diff --git a/hints/linux.sh b/hints/linux.sh index 4e13fd3..b76ee89 100644 --- a/hints/linux.sh +++ b/hints/linux.sh @@ -9,7 +9,11 @@ # Consolidated by Andy Dougherty # # Updated Thu Feb 8 11:56:10 EST 1996 -# Add ability to use command-line overrides for optinal settings. + +# Updated Thu May 30 10:50:22 EDT 1996 by + +# Updated Fri Jun 21 11:07:54 EDT 1996 +# NDBM support for ELF renabled by # perl goes into the /usr tree. See the Filesystem Standard # available via anonymous FTP at tsx-11.mit.edu in @@ -19,7 +23,6 @@ case "$prefix" in '') prefix='/usr' ;; esac -# This may not be needed in 5.002 since sigaction is used. # gcc-2.6.3 defines _G_HAVE_BOOL to 1, but doesn't actually supply bool. ccflags="-Dbool=char -DHAS_BOOL $ccflags" @@ -32,9 +35,6 @@ libswanted="$*" # function in . d_lstat=define -# I think Configure gets this right now, but I'd appreciate reports. -malloctype='void *' - # Explanation? case "$usemymalloc" in '') usemymalloc='n' ;; @@ -64,13 +64,18 @@ main() { } EOM if ${cc:-gcc} try.c >/dev/null 2>&1 && ./a.out; then - cat <<'EOM' + cat <<'EOM' >&4 You appear to have ELF support. I'll try to use it for dynamic loading. +If dynamic loading doesn't work, read hints/linux.sh for further information. EOM - nm_so_opt='-dynamic' + +#For RedHat Linux 3.0.3, you may need to fetch +# ftp://ftp.redhat.com/pub/redhat-3.0.3/i386/updates/RPMS/ld.so-1.7.14-3.i386.rpm +# + else - cat <<'EOM' + cat <<'EOM' >&4 You don't have an ELF gcc. I will use dld if possible. If you are using a version of DLD earlier than 3.2.6, or don't have it at all, you @@ -92,45 +97,67 @@ EOM ## uncomment the next two lines: #ldflags="-static" #so='none' + + # In addition, on some systems there is a problem with perl and NDBM + # which causes AnyDBM and NDBM_File to lock up. This is evidenced + # in the tests as AnyDBM just freezing. Apparently, this only + # happens on a.out systems, so we disable NDBM for all a.out linux + # systems. If someone can suggest a more robust test + # that would be appreciated. + # + # More info: + # Date: Wed, 7 Feb 1996 03:21:04 +0900 + # From: Jeffrey Friedl + # + # I tried compiling with DBM support and sure enough things locked up + # just as advertised. Checking into it, I found that the lockup was + # during the call to dbm_open. Not *in* dbm_open -- but between the call + # to and the jump into. + # + # To make a long story short, making sure that the *.a and *.sa pairs of + # /usr/lib/lib{m,db,gdbm}.{a,sa} + # were perfectly in sync took care of it. + # + # This will generate a harmless Whoa There! message + case "$d_dbm_open" in + '') cat <<'EOM' >&4 + +Disabling ndbm. This will generate a Whoa There message in Configure. +Read hints/linux.sh for further information. +EOM + # You can override this with Configure -Dd_dbm_open + d_dbm_open=undef + ;; + esac fi rm -f try.c a.out if /bin/bash -c exit; then + echo echo You appear to have a working bash. Good. else - cat << 'EOM' -Warning: it would appear you have a defective bash shell installed. This is -likely to give you a failure of op/exec test #5 during the test phase of the -build, Upgrading to a recent version (1.14.4 or later) should fix the -problem. + cat << 'EOM' >&4 +*********************** Warning! ********************* +It would appear you have a defective bash shell installed. This is likely to +give you a failure of op/exec test #5 during the test phase of the build, +Upgrading to a recent version (1.14.4 or later) should fix the problem. +****************************************************** EOM fi -# In addition, on some systems there is a problem with perl and NDBM, which -# causes AnyDBM and NDBM_File to lock up. This is evidenced in the tests as -# AnyDBM just freezing. Currently we disable NDBM for all linux systems. -# If someone can suggest a more robust test, that would be appreciated. -# -# More info: -# Date: Wed, 7 Feb 1996 03:21:04 +0900 -# From: Jeffrey Friedl +# Avoid some troublesome gcvt() functions. With some libc versions, +# perl -e '$x=1e5; print "$x\n";' prints 1e+5. We'd like it +# to print 100000 instead, consistent with the integer value given +# on other platforms. This isn't a bug in gcvt, really; more in our +# expectations for it. We'd like it to behave exactly as +# sprintf %.16g, but it isn't documented to do that. # -# I tried compiling with DBM support and sure enough things locked up -# just as advertised. Checking into it, I found that the lockup was -# during the call to dbm_open. Not *in* dbm_open -- but between the call -# to and the jump into. +# We'll use sprintf() instead, since we can control the output more +# precisely. # -# To make a long story short, making sure that the *.a and *.sa pairs of -# /usr/lib/lib{m,db,gdbm}.{a,sa} -# were perfectly in sync took care of it. -# -# This will generate a harmless message: -# Hmm...You had some extra variables I don't know about...I'll try to keep 'em. -# Propagating recommended variable d_dbm_open -case "$d_dbm_open" in -'') d_dbm_open=undef ;; -esac +# The next version of Configure will check for this automatically. +d_Gconvert='sprintf((b),"%.*g",(n),(x))'