rpm4/rpm5 incompat: rpm5 firing on provides

[Jeffrey Johnson]

On Oct 30, 2012, at 3:59 AM, Jan Rękorajski <baggins at pld-linux.org> wrote:

> On Mon, 29 Oct 2012, Elan Ruusamäe wrote:
> 
>> # rpm -q rpm
>> rpm-5.4.10-18.x86_64
>> 
>> # rpm -q rc-scripts initscripts
>> rc-scripts-0.4.3.3-3.x86_64
>> rc-scripts-0.4.5.4-2.x86_64
>> package initscripts is not installed
>> 
>> # rpm -q --whatprovides initscripts
>> rc-scripts-0.4.3.3-3.x86_64
>> rc-scripts-0.4.5.4-2.x86_64
>> 
>> # rpm -e rc-scripts-0.4.3.3-3.x86_64
>> mv: cannot stat ‘/etc/sysconfig/network-scripts/ifcfg-*’: No such file 
>> or directory
>> error: %trigger(rc-scripts-0.4.3.3-3.x86_64) scriptlet failed, exit status 1
>> mv: cannot stat ‘/etc/sysconfig/network-scripts/ifcfg-*’: No such file 
>> or directory
>> error: %trigger(rc-scripts-0.4.3.3-3.x86_64) scriptlet failed, exit status 1
>> 
>> # rpm -q --triggers rc-scripts-0.4.3.3-3.x86_64
>> triggerpostun scriptlet (using /bin/sh) -- initscripts
>> mv -f /etc/sysconfig/network-scripts/ifcfg-* /etc/sysconfig/interfaces
> 
> Can you provide output of 'rpm -vvv' on rpm4 and rpm5?
> 

Before we even get to -vv output, I have these questions:

1) You have 2 versions of rd-scripts installed:
	Is that "normal" or already buggy?

2) The packages are cross-triggering on provided "initscripts"
	Is that "normal" or just ancient hysteria that noon
	wants to fix?

3) A triggerpostun runs after package removal
	Is there really any surprise why the triggerpostun is failing?
What isn't triggerpreun (or hardening of the triggerpostun script) being
attempted?

> Jeff, will changing:
> 	rpmdsNew(sourceH, RPMTAG_PROVIDENAME, 0);
> to:
> 	rpmdsNew(sourceH, RPMTAG_NVRA, 0);
> 
> in handleOneTrigger in lib/psm.c:~1338, give the old behavior?

I'm not sure what "old behavior" means here: I don't use PLD.

(aside)
At some point I also lose interest in helping revert to changes
made in RPM 5+ years ago: you could in fact just continue running
same old rpm-4.5 forevermore rather than screwing around
with rpm-5.x.y behavior to conform to your expectations (and
making it much harder to "support" all the patched/reverted/altered
behaviors.

But sure s/PROVIDENAME/NVRA/ is a start towards using
package names rather than provide names for triggers.

I believe you will find other surprises …


73 de Jeff

> 

> -- 
> Jan Rękorajski                                 | PLD/Linux
> SysAdm                                         | http://www.pld-linux.org/
> baggins<at>mimuw.edu.pl
> baggins<at>pld-linux.org
> _______________________________________________
> pld-devel-en mailing list
> pld-devel-en at lists.pld-linux.org
> http://lists.pld-linux.org/mailman/listinfo/pld-devel-en