False positive dependency checks break package install order

Jeffrey Johnson n3npq at me.com
Tue Aug 13 18:26:38 CEST 2013

On Aug 13, 2013, at 12:15 PM, Mariusz Mazur wrote:

> On Tue of August 13 2013, Jeffrey Johnson wrote:
>> Meanwhile the simpler fix is to adjust the Bloom filter parameters to
>> decrease the probability of a false positive.
> Are you serious? You really think it's ok for rpm to be using algorithms with 
> a non-zero chance of false positives?

Yes. The Bloom filters been in use for about 3 years with no known problems,
including randomized installations for 5-10 linux distros under a CI harness.

THe performance/memory gains are significant.

> Are there any other places in the code with algos like this?

Yes. Bloom filters are quite convenient.

73 de Jeff

More information about the pld-devel-en mailing list