RE: Generating C# from Java (Was: fit, nfit, et al?)

Ward,

I recall your describing how you came up with Fit. It sounded like a rather
chaotic process, involving false starts, experimentation and change. I
wouldn't be in favor of a process that made such things less likely to
happen in the future.

IMO, the .NET version should look as .NET code generally looks when written
by people immersed in the .NET world. The same goes for Python, Ruby, etc.
The problem with C# is that it looks enough like Java to make it appear
possible to keep the code the same. Well, it may be possible, but I 
suspect it ain't desirable.

For example, most people I know camel-case fields but not properties or
methods in C#. That's a small distinction, but anyone who is knowledgable
enough to be going back and forth between the Java and C# versions will
probably adjust pretty readily.

More important, I think that different language versions may need to have
different capabilities entirely. And I agree with Steve that the ferment
of ideas that comes up with different folks working on different versions
is likely to produce exciting advances for all versions.

What I think is needed - to keep all of this still FIT - is a set of 
acceptance tests that all implementations must pass.

Charlie

Charlie Poole
[email protected]
http://www.pooleconsulting.com
http://www.charliepoole.org


> -----Original Message-----
> From: [email protected]
> [mailto:[email protected]]On Behalf Of Ward Cunningham
> Sent: Friday, August 08, 2003 9:44 AM
> To: FIT Developers
> Subject: Re: [Fit-dev] Generating C# from Java (Was: fit, nfit, et al?)
> 
> 
> 
> On Thursday, August 7, 2003, at 11:59  PM, Jim Little wrote:
> 
> > I haven't talked to Ward about this yet so I don't know what he thinks.
> > When we first started the port, we looked at a regex-style approach but
> > postponed it in favor of getting a first release out, and we never got
> > back to it.
> 
> Also, when we wrote the c# we made a few name changes required by c# 
> with the best intention of moving those name changes back into the java 
> just so they would look the same to someone going back and forth 
> between c# and java.
> 
> Regretfully those name changes were never made. Had we used the 
> generator approach we never would have had two sources that could get 
> out of sync.
> 
> As I write this, I am wondering how simple I could make a generator If 
> I wrote two generators, one for java and one for c#, that both ran off 
> of the same "generator specification". Snobol was written that way, as 
> all "macro" calls. They must have used something on the level of m4. I 
> bet it would be cake to write a macro processor that was even more 
> suitable for object code than m4 given that we have things like perl, 
> python or ruby for the macro implementation language.
> 
> Of course generating by machine or by hand would be easier if we had 
> good acceptance tests. At fit-fest next week my own goal will be to 
> advance the fit acceptance tests (fat) which is a complementary way to 
> enforce consistency between implementations.
> 
> --
> Ward Cunningham
> 503-245-5633 v	mailto:[email protected]
> 503-246-5587 f	http://c2.com
> 
> 
> _______________________________________________
> Fit-dev mailing list
> [email protected]
> http://lists.freepan.org/mailman/listinfo.cgi/fit-dev