Re: BUG #7823: PGDG packaged pgadmin3_92 removes its /usr/bin/pgadmin3 link - Mailing list pgsql-bugs

From Georgi Georgiev
Subject Re: BUG #7823: PGDG packaged pgadmin3_92 removes its /usr/bin/pgadmin3 link
Date
Msg-id 20130203043654.GF14929@jnext-0060.corp.japannext.co.jp
Whole thread Raw
In response to Re: BUG #7823: PGDG packaged pgadmin3_92 removes its /usr/bin/pgadmin3 link  (Georgi Georgiev <georgi-georgiev-pgsql@japannext.co.jp>)
Responses Re: BUG #7823: PGDG packaged pgadmin3_92 removes its /usr/bin/pgadmin3 link
List pgsql-bugs
Quoting Georgi Georgiev at 03/02/2013-13:19:52(+0900):
> Quoting Georgi Georgiev at 24/01/2013-12:55:31(+0900):
> > Quoting Devrim G=DCND=DCZ at 23/01/2013-21:11:09(+0200):
> ...
> > >=20
> > > Should be fixed with 1.16.1-2. Please check.
> >=20
> > Thanks for the quick turnaround. But there seems to be a typo now, so
> > the %post is not working:
> ....
>=20
> 1.16.1-3 does not throw errors, but it still reliably wipes out the
> symlink on an upgrade. A fresh install or a reinstall would create the
> symlink, but that is only because the %preun does not run in that case
> (there is no old version to "uninstall" in this case).
>=20
> This can be fixed with a 'if [ "$1" =3D 0 ]; then' in the %preun.
>=20
> To test all these cases, I used rpmrebuild to create different releases=
 of this
> package.
>=20
> I made a "-2.5" which is identical to -3.
> And I made a "-2.8" with the %preun modified like this:
>=20
> $ rpm -q pgadmin3_92-1.16.1-2.8.rhel6.x86_64  --scripts
> postinstall scriptlet (using /bin/sh):
> /usr/sbin/update-alternatives --install /usr/bin/pgadmin3 pgadmin3 /usr=
/pgsql-9.2/bin/pgadmin3 920
> preuninstall scriptlet (using /bin/sh):
> if [ "$1" =3D 0 ]; then
>   /usr/sbin/update-alternatives --remove pgadmin3 /usr/pgsql-9.2/bin/pg=
admin3
> fi
>=20
> Test results:
>=20
> - Upgrade from 2.5 - symlink missing (wrong)
> - Upgrade from 2.8 - symlink present (correct)
> - Fresh install of any version - symlink present (correct)
> - Uninstall any version - symlink missing (correct)
>=20
> So, please, if you can fix the %preun as above, that should solve all p=
ending
> issues.

By the way, I see that the postgresql92 packages do the
"update-alternatives" cleanup in "%postun". I am not sure why
pgadmin3_92 does it in "%preun", nor does it seem to make much of a
difference, but it may be a good idea to consistently do it in the same
scriptlet - either %preun or %postun.

--=20
(    Georgi Georgiev   (  Despite all appearances, your boss is a    (
 )    Sysadmin Head     ) thinking, feeling, human being.             )
(    -SBI Japannext-   (                                             (

pgsql-bugs by date:

Previous
From: Maciej Łopuszański
Date:
Subject: Re: BUG #7842: pg_dump scripts view to table
Next
From: Kevin Grittner
Date:
Subject: Re: BUG #7846: Documentation: Should FOUND be documented as an output of UPDATE statement?