On Mon, 2021-02-08 at 12:21 -0600, Alex Stewart wrote: > On 2/2/21 5:48 PM, Richard Purdie wrote: > > A build date was leaking into the generated docs and makefile used for > > ptests leading to reproducibility issues each time the month changed. > > > > Add a patch to use SOURCE_DATE_EPOCH to derive it if available. > > ACK the objective, but I have a couple concerns about the patch below. > > > +--- opkg-0.4.4.orig/configure.ac > > ++++ opkg-0.4.4/configure.ac > > +@@ -281,7 +281,12 @@ AC_FUNC_UTIME_NULL > > + AC_FUNC_VPRINTF > > + AC_CHECK_FUNCS([memmove memset mkdir regcomp strchr strcspn strdup > > strerror strndup strrchr strstr strtol strtoul sysinfo utime]) > > + > > +-CLEAN_DATE=`date +"%B %Y" | tr -d '\n'` > > ++1607446883 > > Is this `1607...` timestamp just some leftover pollution in the > configure.ac file or does it serve some purpose?
Sorry, its just junk that crept in from testing. I'm surprised that doesn't error! > > ++if [ ! -z "$SOURCE_DATE_EPOCH" ]; then > > AutoConf doesn't have the shell interpretation of `[`; right? At least > on my system, AutoConf drops the brackets in this statement and writes > the ./configure file as: > > if ! -z "$SOURCE_DATE_EPOCH" ; then > > which obviously fails when executed. Can you use `if test ! -z ...` instead? I should know better than using [] in configure :) The other format should be fine. It probably works as OE always sets this now. Cheers, Richard
-=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#147837): https://lists.openembedded.org/g/openembedded-core/message/147837 Mute This Topic: https://lists.openembedded.org/mt/80341753/21656 Group Owner: openembedded-core+ow...@lists.openembedded.org Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub [arch...@mail-archive.com] -=-=-=-=-=-=-=-=-=-=-=-