On Mar 31, 2008, at 3:28 PM, Jeff Johnson wrote:

>
> On Mar 29, 2008, at 12:40 PM, devzero2000 wrote:
>> I have tried this on rpm 5.0.3 but it fail
>>
>> https://bugzilla.redhat.com/show_bug.cgi?id=205350
>>
>>
>> e.g.
>>
>> rpm -e R R-mvtnorm
>> /var/tmp/rpm-tmp.48546: line 1: /usr/lib/rpm/R-make-search- 
>> index.sh: No such file or directory
>> error: %postun(R-mvtnorm-0.8-2.el5.i386) scriptlet failed, exit  
>> status 127
>> [root@handrax tmp]# rpm -v --version
>> rpm (RPM) 5.0.3
>> rpmlib 0x28100c00,0x7d831800,0x52504d35
>>
>> What's happen ? It was fixed, iirc, some time ago also in rpm  
>> 4.4.2.-x
>>
>
> (I was travelimg yesterday and the 20+ Mb of the R package was
> more than GPRS on my cell could handle).
>
> The issue of whether there is a regression is different than the issue
> of whether there is a problem.
>
> I have a reproducer now, will have a fix in a day or so.
>

Should be fixed, at least the reported R R-mvtnorm erasure is invariant
to the order mentioned on the CLI.

In both orders, R-mvtnorm is erased before R, and so %postun runs  
successfully.

I'll have to live on the change for a week or so to see if anything  
else broke.

Thanks for reporting.

73 de Jeff