Description of problem: This patch defines ISANAME and ISABITS so that it makes rpmbuild stop the following message. > error: Macro %__isa_name has empty body > error: Macro %__isa_bits has empty body It also needs for rpmbuild to produce debuginfo package. --- rpm-4.6.0-rc1/installplatform.org 2008-10-21 14:46:48.000000000 +0900 +++ rpm-4.6.0-rc1/installplatform 2008-10-22 10:00:22.000000000 +0900 @@ -116,6 +116,10 @@ ISANAME=ia ISABITS=64 ;; + sh*) + ISANAME=sh + ISABITS=32 + ;; esac if [ -n "$MULTILIBNO" ]; then I also revised patch for rpm-4.6.0-rc1 which didn't include in Bug #432496. Version-Release number of selected component (if applicable): rpm-4.6.0-0.rc1.2
Created attachment 321118 [details] rpm-4.6.0-sh-linux.patch add option to optflag.
Created attachment 321119 [details] rpm-4.6.0-shisa.patch
Both patches are now applied in upstream, thanks for them. Just a quick question. Since I have no access to SH arches, what problems do you see when the IEEE compliance flag is missing from CFLAGS?
(In reply to comment #3) > Both patches are now applied in upstream, thanks for them. > > Just a quick question. Since I have no access to SH arches, what problems do > you see when the IEEE compliance flag is missing from CFLAGS? I found the perl and mpfr package didn't pass self test without it. After that, I alwasys build packages using patched rpm ,so I don't know other packages which really need this option.
Here are build log which I had found. a part of mpfr-2.3.0-1.fc8 's build log PASS: tset_exp PASS: tset PASS: tabs tset_d.c:76: assertion failed: (!(((d) >= 0.0) + ((d) <= 0.0))) /bin/sh: line 4: 1005 Aborted MPFR_QUIET=1 ${dir}$tst FAIL: tset_d PASS: tset_f ..... PASS: tcmp2 PASS: tcmpabs /bin/sh: line 4: 1306 Terminated MPFR_QUIET=1 ${dir}$tst FAIL: tcmp_d PASS: tcmp_ld a part of perl-5.8.8-30.fc8 's build log t/op/chr..................................ok t/op/closure..............................ok t/op/cmp..................................FAILED at test 46 t/op/concat...............................ok t/op/cond.................................ok
Seems like a good reason to add it then. Thanks for the patches again :)