Test coverage for a compiler is one thing: the compiler is fully defined.

Test coverage for a large app is an entirely different matter. When you've got 500+ forms, etc., even Java or .Net aren't going to help you. This is not an original thought, btw, it is echoed in multiple language development settings.

OTOH, running a web service model does make testing quite easy. With Python, especially, testing is just there, with very little work required to implement and manage it. If Plan A (pyjamas/dabo) doesn't fly for any reason, and Lianja does fly, we would be doing our work in Python (which is to day we will be using Python either way). I have no objection at all to writing in VFP (doing it right now, in fact). And I get additional enjoyment in using Python, much in the same way that eating with dinnerware that matches one's aesthetic taste and functional requirements brings additional pleasure to food.

Hank

>>Lianja (http://www.lianja.com) promises to be the killer developer system for which we have all been waiting.
>>
>>-- VFP compatible
>>-- programmable in Recital Script, VFP, Python, Javascript (they obviously dream of Abstract Syntax Trees <s>)
>>-- hit DBF's (they are a project within Recital, which has been doing xBase forever) or SQL
>>-- WebClient automatically generated, runnable on any platform, any device
>>
>>There are many questions to be answered, of course:
>>
>>-- will it really do all that?
>>-- will they be responsive to developer needs?
>>-- will the cost be feasible for both small custom development shops and ISVs with broad deployment?
>
>How much test coverage will they have with each language, device, backend setup ?
>I guess they moved in the last few years in big/middle iron circles.
>Could be too much - the large apps I was involved in grew static because of test overhead,
>even when implementing partially automatic testing. And that was in places they were used to support
>a couple of thousand users for each app in a laaaarge suite.
>
>[One of the reasons some went to java - much higher dev cost, but less test cost]
>
>>If they deliver, I can't imagine a much better tool to move from today into the future.
>
>looks nice on the site, they probably will finish, but without listed product licensing scheme
>and at least an estimated price range I won't look at the beta.
>
>And vfp devs are among the pickiest bunch of an already strange craft (me definately included...)
>
>regards
>
>thomas