On Jan 28, 2008, at 12:59 PM, Jeff Johnson wrote:

>
> The real fixing is deeper, enriching the return codes from  
> rpmtsCheck() to
> reflect whether problems were found.
>
> I've attached a patch that fixes your immediate problem, but I need  
> to check
> a few more code paths, and I may return rpmRC instead of {0, 1, 2}  
> integers.
>

Here's the nest evolutionary patch to permit 0/1/2 return codes from  
rpmtsCheck.

My next attempt will use RPMRC_OK/RPMRC_NOTFOUND/RPMRC_FAIL
instead of 0/1/2 as return codes.

I may also attempt to parameterize whether rpmtsCheck() returns with  
1st failure,
or with all failures, included in problem sets. Returning 1st  
failure, as well as skipping
"known good" checks,  are quite important speed-up if not using the  
rpmtsCheck() callback.

73 de Jeff