Glenn Fowler wrote: > some of the ast header setup is from the 80's when many > /usr/include headers couldn't be trusted > I have a cleanup on my todo list > send me a diff -r (offlist) between the 32 and 64 bit headers > and I'll add that to the cleanup
I attached the diff already to my previous email as "ksh93_sparc_32bit_arch_vs_64bit_arch.diff.txt.gz". http://mail.opensolaris.org/pipermail/ksh93-integration-discuss/2006-March/000155.html has the link (http://mail.opensolaris.org/pipermail/ksh93-integration-discuss/attachments/20060330/7e22e467/ksh93_sparc_32bit_arch_vs_64bit_arch.diff.txt.bin) to the gzip-compressed file. Getting the diffs for "i386 vs. AMD64" may be slightly more problematic to obtain as I do not have access to a AMD64 machine from home... ;-( > but please note that some of the headers are iffe generated > (moral equivalent of config.h) so there may be > valid reasons for the differences that opensolaris may > have to mitigate via #ifdef solaris_A ... #ifdef solaris_Z > > unless absolutely necessary we will not code > #if system_type_foo_VERSION < 1234 > and instead will continue to do feature based conditionals > > otherwise the code will become a maintenace nightmare > perpetually steps behind system_type_foo releases Agreed. I was just thinking about having one uniform version of the includes for one specific OS version (such as "Solaris 10") which should simply be independent of the hardware (e.g. 32bit/64bit on both x86 and SPARC (=four combinations)). IMHO it would be tricky to convince Sun to have multiple versions of the include files around (at least in the "32bit vs. 64bit case", it may be valid to have different sets of include files for "x86 vs. SPARC") ... ---- Bye, Roland -- __ . . __ (o.\ \/ /.o) roland.mainz at nrubsig.org \__\/\/__/ MPEG specialist, C&&JAVA&&Sun&&Unix programmer /O /==\ O\ TEL +49 641 7950090 (;O/ \/ \O;)
