On Nov 18, 2007, at 11:04 AM, Jeff Johnson wrote:

>
>>
>> Notice the "error: rpmdbNextIterator: [...]" error lines!
>> Jeff, any clue?
>
> The value of ril in this call to headerVerifyInfo (near pkgio.c: 
> 956) has rotted:
>
>     /* If no header-only digest/signature, then do simple sanity  
> check. */
>     if (info->tag == 0) {
>         xx = headerVerifyInfo(ril-1, dl, pe+1, &entry->info, 0);
>
> I should have a fix in the next couple of hours.
>

Symptom fix in a moment, real fix as soon as I can figger.

Really sick and twisted, been b0rked in rpm forever.

Turns out that pubkey headers are _NEVER_ sanity checked.
Instead, the program flow silently _DISABLED_ the header tag
sanity check.

When I changed the 1st arg of headerVerifyInfo to unsigned, the loop  
across
tags actually became active, rather than being disabled.

So the symptom fix is to not attempt header tag sanity checks for
pubkeys, restoring behavior status quo ante.

I'll attempt a real fix this week. It's quite ironic that pubkey headers
are in RPMv3 format, have incorrect data types for Summary/ 
Description/Group,
are never sanity checked, and have almost nothing to do with "package
management" per se.

And all these bugs are showing up while I'm eliminating support for  
RPMv3
headers.

73 de Jeff