Method PageRank for Reflector

My blog has moved.
You can view this post at the following address:
http://www.osherove.com/blog/2004/6/4/method-pagerank-for-reflector.html
Published Friday, June 04, 2004 11:53 PM by RoyOsherove
Filed under:

Comments

Friday, June 04, 2004 8:25 PM by Tom

# re: Method PageRank for Reflector

Agree. This is really cool:
http://blog.dotnetwiki.org/archive/2004/06/04/311.aspx
Sunday, June 06, 2004 5:02 PM by Jonathan de Halleux

# re: Method PageRank for Reflector

:)

You can play with the tool, it's online on my blog :)
Monday, June 07, 2004 10:40 AM by Barnea Gal

# re: Method PageRank for Reflector

I like it and think it's great for a getting this kind of info quickly, but ...

I believe the using a code profiler (I use ants ...) and analizing the results on an actual run, will give a better understanding of which functions are the most important to debug (called the most times, take longest to process, etc ...).

Don't get me wrong, i'm not comparing one-man's work to a known company's product , but monitoring an actual run of a program will give you (i think) better results




Friday, June 11, 2004 2:04 AM by Jonathan de Halleux

# re: Method PageRank for Reflector

Hi Barnea,

MethodRank and Ants have one big difference: Ants needs you to execute the code in order to record which part is run, how many time, ... this is "runtime" analysis. MethodRank does not need to run the code and it's analysis is only based on the metadata contained in the assembly, this is "static" analysis. Both approach have advantages and caveats, based on their intrisinsic nature:

- a profiler needs you to run an application: how long should it run ? how do you run it ? are you sure you use all the funcionalities in the test run ? etc...
- methodRank is based on a "BIG" assumption!

While Ants give you the "what", MethodRank gives you a hint (and not more than that) on the why. :)

I have not had the opportunity to see the correlation between a profiler result and Ants, would be interresting.