[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: Testing libsatsolver on Fedora



On 07/31/2009 03:46 PM, James Antill wrote:

>  We also don't make progress by posting "yum is 50x slower than solv for
> update", and yet _predictably_ that is what this thread degenerated into
> within hours of your post.

I seem to be reading a different thread.  A few people were trying out
the code posted and posting feedback and rough time results.

>  Then why post the numbers, if you know they aren't
> comparable/worthwhile/etc?

Are you familiar with the concept of a 'smell test'?  Here:
  http://en.wiktionary.org/wiki/smell_test

   1. (idiomatic) An informal method for determining whether
   something is authentic, credible, or ethical, by using one's
   common sense or sense of propriety.

The results of a time on 'solv update' vs 'yum update' aren't
comparable, but they are worthwhile.  Michael had posted that the SAT
algorithm can be faster and more accurate.  That's a large claim.  The
SUSE wiki claims it can be magnitudes faster ("but than what," was
validly raised).  My smell test showed that the two achieved similar
results within a few multiples.  That narrows the claims - solv isn't
magnitudes faster as presently composed and I stated I didn't know
exactly how much less work it's doing than yum.  We also don't know how
well it's been optimized.  The results show that it's at least possible
that it could perform faster - it's certainly not magnitudes slower.
So, it passes the smell test - the claim isn't complete BS.  If it were
a hundred times slower it wouldn't pass the smell test.

Nobody pretends a smell test is a rigorous benchmark.  In fact, spending
time on rigorous benchmarking without first performing smell tests would
be foolhardy.

>  And, yeh, I'd have at least worded my reply differently (if not just
> hit delete thread) if you were the first person to ever post weird
> numbers and call it a yum vs. BLAH benchmark. But that works the other
> way around too.

I think if you'll look again you'll see that I did not call it a
benchmark nor infer that it was.  You're making a strawman argument
here.  Apparently some people are unkind about yum's performance but you
can't assume that anybody looking at yum performance is mounting an attack.

-Bill
-- 
Bill McGonigle, Owner           Work: 603.448.4440
BFC Computing, LLC              Home: 603.448.1668
http://www.bfccomputing.com/    Cell: 603.252.2606
Twitter, etc.: bill_mcgonigle   Page: 603.442.1833
Email, IM, VOIP: bill bfccomputing com
Blog: http://blog.bfccomputing.com/
VCard: http://bfccomputing.com/vcard/bill.vcf


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]