i'm really no JS expert, also JS functionality cannot be depended on (we're dealing with constrained browsers here), so here's a list of issues i ran into and remarks about them. please add to the list as needed (this calls for a wiki...).


still to implement:
- rewrite of the capabilities test page currently available under http://pwmwa.com/frost
	- plan is to have a "cascaded test"
	- this means we first define all the features a browser needs to support in order to run the Frost library 
		this includes basic JS support, XHR, innerHTML etc. 
		cascaded means that the page loads and then executes one test after another automatically 
		if that's possible (controlled by JavaScript)
		if a test fails and the cascade breaks, a link will be shown that the user has to click. 
		so in the ideal case the user has nothing else to do than to open a URL and after a while is presented with a page saying "thank you, your browser passed all tests".
- definition of WURFL properties for tested capabilities
- export of capabilities into WURFL patch file like this:
<wurfl_patch>
	<devices>
		<!-- for all tested browsers -->
		<device user_agent="Frost_tested_UA" fall_back="generic" id="frost_tested_ua">
			<group id="ajax">
				<capability name="xhr_native" value="false"/>
				<capability name="xhr_activex" value="Msxml2.XMLHTTP.5.0"/>
				<!-- all other tested capabilities -->
				<capability name="frost_lib_supported" value="false|basic|full"/>
			</group>
		</device>
		<!-- endfor -->
	</devices>
</wurfl_patch>
- remember: declare at the 'generic' (unrecognized) device level that those properties are false


always an issue:
- size savings in code


object-orientize everything
- useful or not?
- size overheads?
- e.g.: frost.toid(['this text', 'here']);


f() args passing as JSON objects?
- need to test for eval() availability
- maybe size overhead because of {name:value} vs. [value]


f() args default values and passing of empty values best practices
- now passing empty parameters this looks like this: ['value', 'value2', '', '', 'value5'] 
- defaults assignments like this: var meth = arg[2] != '' ? arg[2] : 'POST';


is "typeof" trustworthy?
- anybody?

flashing of divs
- internet explorer has a problem with setTimeout() - calling it sends form
- doing it in a while loop delays script execution by while-duration