Inspired by CRC Cards, PatternCards captures preliminary patterns that make PatternLanguages.

I am developing a fairly complex pattern language with another person and we need to collect and arrange the patterns we have discovered.  We are using index cards to help organize the patterns and visualize the language.

Here are the steps:

	* Write a Pattern name at the top of an index card.
	* Write a short description of the problem.
	* Write a solution (if you have already identified one).
	* As you develop more cards, lay them out on the table and see if connections arise.
	* Patterns that are subordinate to another should have its ''super-pattern'' name written at the top right corner of the card.
	* Related (subordinate or peer) pattern names are written along the bottom of the card.
	* The relationship among the patterns are discussed and networks begin to emerge...
	* Make sure each card really represents a pattern. 
	* Once you are sure of the patterns you got, break them up (by network) and start expanding the cards into ''real'' patterns.

This is working wonderfully so far.  I can't be the only one who has seen the applicability of CRC-like techniques to
pattern discovery.  Anyone else have experiences with developing pattern languages this way? 

-- ToddCoram
-------------------------------------------
This reflects my experience as well. I'm not sure I use "Patterns" in quite the same way as the rest of you, though.

I've always viewed classes & methods as something like instances of PrimitivePattern, though.

So, from my perspective, its only reasonable that PatternCards approach would generalize from CRC Cards. Kudos for recognizing it!

I think the hallmark of a rich and useful idea is the feeling, after the fact, that it should have been obvious all along. Kudos on the discovery of an important extension to pattern methodology.

-- TomStambaugh
-------------------------------------------
I've also used CRC cards to work on pattern languages,
in particular, the FoundObjects language which developed
out of the ResultObjectPattern.

My cards are somewhat simpler than those described 
above, more like the basic CRC cards.
Each of my cards includes a name at the top, and (being a traditionalist)
related patterns down the right hand side.
The body of the card contains a PatLet, WardCunningham's stylised
version of a problem-solution pair, separated by the word
ThereFore in BiroBold.

I found these cards very helpful for organising a pattern 
language, but less so for finding patterns in the first place.

-- JamesNoble
-------------------------------------------
Describe your experiences here.