RE: chunked output
"Charlie Poole" <[email protected]> Tue, 19 Aug 2003 11:53:55 -0700
| Newsgroups | gmane.comp.programming.tools.fit.devel |
|---|---|
| Message-ID | <[email protected]> |
Ward, > > By using chunked encoding I can make FitNesse do this. But it means > > that > > FIT will need to spit out the results as soon as it makes them. FIT > > needs > > chunked output. > > In a previous fit-like testing framework I found it advantageous to > process tables in other than sequential order. This was due to a quirk > in the way the domain specialist chose to organize tables in her > spreadsheet. If we find a good way to allow subclasses of the top level > Fixture then we could see this again in Fit. A top level fixture could > override doTables and do them any way it wanted. > > I'm guessing by chunking Micah means that we write annotated html for > each table as it is completed. This would be only one kind of progress > indication that would be possible. Another would be some sort of event > or callback as each cell is processed or each annotation is made. In > java I would do this with some sort of Listener interface. What would > other implementations do? In C# you could use an interface or have the fixture provide an event to be subscribed to. You could do the analog of either in many other other languages, but maybe not all. This seems to introduce a third interface to those used by test writers and fixture developers: the host application interface. To what extent do we believe this must be standard across all languages? Could we simply provice certain functionality and require the host to write the glue for the various implementations? Or does it have to look exactly the same to the host out of the box? Charlie Poole [email protected] www.pooleconsulting.com www.charliepoole.org