[Windmill-dev] asserts and waits

[Adam Christian]

I forgot to reply to the list, and that didn't get cached in sent  
mail, could you reply all with my response.

Adam

On Aug 29, 2007, at 1:50 PM, Mikeal Rogers wrote:

>> So for each element like what you are talking about, we will need  
>> to add a registry for the UI to dynamically build. If our main  
>> concern is that the UI is going to have 50 asserts and 50 waits I  
>> can understand why this is a little silly, especially when we get  
>> huge amounts of other functions. ) we would have to add specific  
>> code in there to say is it an assert, then split the string to  
>> find out which one, or for wait, do the same thing.. I have a  
>> feeling there is going to be more sets of things that share a  
>> common prefix.
>
> Let me get one thing straight. The windmill controller API is a  
> public API that currently only translates in to 3 usages [ the  
> python API, the JSON api, and the IDE UI ].
>
> Moving forward that API end up being used by a variety of other  
> language and library developments in the community. In order to  
> make this easy for people it needs to be simple, clean, and  
> standard across all the usages. So fixing the problem by  
> consolidating some of the public API functions in the IDE doesn't  
> help us.
>
> Also, this doesn't effect the extensions. The extensions will  
> define new assertions as new controller functions, so special  
> casing the IDE UI could be built statically instead of dynamic  
> since the only code for it will be in the core windmill API.
>
>> We want to have assertText be one function in the API (which for  
>> maintainability, we need each of these to be their own function
>
> Why is this more maintainable? The public API, documentation, and  
> implementation by other languages sure isn't more maintainable this  
> way.
>
> The proposal is for the _public_ assert to be consolidated. Your  
> actual implementation doesn't have to be one big monolithic  
> function, you can dispatch to a bunch of other functions called  
> assertText, assertBlah, that's up to you. I imagine for code  
> maintainability you'll definitely want to do this. But the public  
> API should be consolidated and standardized.
>
> asssert(type={identifier object}, value=obj)
>
> We can default value to be "exists", since in more cases you'll  
> just want to assert than some identifier exists.
>
> -Mikeal
> _______________________________________________
> Windmill-dev mailing list
> Windmill-dev at osafoundation.org
> http://lists.osafoundation.org/mailman/listinfo/windmill-dev