Yes, and I have not had time to understand and test it. I have not committed
anything because I don't want to make it worse. For example your latest
patch does something special for the "etc" set, and I don't understand how "etc"
is special. Also obsolete_dir is only used in one part of the if statement and
should not be always set.
christos
> On Dec 21, 2019, at 9:20 PM, Izumi Tsutsui <tsutsui%ceres.dti.ne.jp@localhost> wrote:
>
> The following reply was made to PR bin/54730; it has been noted by GNATS.
>
> From: Izumi Tsutsui <tsutsui%ceres.dti.ne.jp@localhost>
> To: christos%netbsd.org@localhost
> Cc: gnats-bugs%netbsd.org@localhost, tsutsui%ceres.dti.ne.jp@localhost
> Subject: Re: PR/54730 CVS commit: src/usr.sbin/postinstall
> Date: Sun, 22 Dec 2019 11:15:46 +0900
>
> I wrote:
>
>> /var/db/obsolete/etc is not extracted into ${DESTDIR}
>> as other sets but prepared in ${ETCTGZDIR}:
>>
>> --- postinstall.in.orig 2019-12-04 20:36:27.562323424 +0900
>> +++ postinstall.in 2019-12-04 20:37:31.591546076 +0900
>> @@ -1465,11 +1465,18 @@
>> # $3: setname
>> select_obsolete_files()
>> {
>> + local obsolete_dir
>> + if [ $3 = "etc" ] ;then
>> + obsolete_dir=${SRC_DIR}/var/db/obsolete
>> + else
>> + obsolete_dir=${DEST_DIR}/var/db/obsolete
>> + fi
>> +
>> if $SOURCEMODE; then
>> ${SED} -n -e "/obsolete/s@\.$1$2[[:space:]].*@\1@p" \
>> ${SRC_DIR}/distrib/sets/lists/$3/mi
>> else
>> - ${SED} -n -e "s@\.$1$2\$@\1@p" "${DEST_DIR}/var/db/obsolete/$3"
>> + ${SED} -n -e "s@\.$1$2\$@\1@p" "${obsolete_dir}/$3"
>> fi
>> }
>>
>>
>
> You have a concern (in PR/54741):
>
> https://mail-index.netbsd.org/netbsd-bugs/2019/12/07/msg064879.html
>>> I am worried that SRC_DIR does not work if not $SOURCEMODE, this is why
>>> I have not committed it...
>
> My answer is:
>
> https://mail-index.netbsd.org/netbsd-bugs/2019/12/07/msg064883.html
>>> I guess !$SOURCEMODE means $TGZMODE.
>>> In that case SRC_DIR for the etc set is already handled,
>>> at least it should work in select_obsolete_files():
>
> Do you have some idea about specific test cases?
> These postinstall problems should be fixed before 9.0 release.
>
> Thanks,
>
> ---
> Izumi Tsutsui
>
Attachment:
signature.asc
Description: Message signed with OpenPGP