[antlr-interest] antlr & antlr studio...

[Hill, Robert]

That's fantastic jim, - just what I wanted to hear :)
I think I'll dip my toes in the water, and use the java code until the
c# codegen is done, they're not that different anyway I suppose, its
just to enviroment I was after ;)
Im not bothered by the early access, I've been used to worse hehe! And
the lack of docs is a bit of a pain, but I've been following the
blog's/Ter's mail's so I've got a reasonable idea what the differences
are, so along with the examples I think I'll be ok :)

Thanks!

Rob



--
Rob Hill
EDS - Hallamshire Business Park
F1E/087
Sheffield	
T:	+44 (0) 114 291 1928
M:	+44 (0) 791 732 1227
E:	rhill03 at eds.com

 

>-----Original Message-----
>From: Jim Idle [mailto:jimi at temporal-wave.com] 
>Sent: 23 April 2006 09:20
>To: Hill, Robert; antlr-interest at antlr.org
>Subject: RE: [antlr-interest] antlr & antlr studio...
>
>> Out of interest , how stable is V3?  Does it do everything 
>it says on 
>> the tin yet? :)
> 
>I find that you need the occasional work around but that 
>because of the nature of it, if you have produced somethign 
>that works once, it will always work. The lack of documenation 
>at the moment is probably the most difficult thing to surmount 
>but you can piece it all together from teh examples if you are willing.
>
>> However when it hits VS2005, I'll almost certainly use your 
>C codegen 
>> since at least theres a modicum of interop between that and 
>C#..hmm I 
>> wonder if turning on the managed C++ for it would be an 
>option in that 
>> case?
>
>I shall look into this as well as generating a COM interface 
>for the generated code too as plenty of people still need to 
>use this. It might make more sense to use the C++ output for 
>such things but in theory it should all be the same under the 
>hood. Further of course, there is a C# output in in progress 
>(not sure how far along it is), which obviously make the most 
>sense if you are accessing the generated code from C# unless 
>there were some performance penalty. However the generated 
>code is logically efficent so it is only the choices made with 
>the runtime implementation that would have dramatic effects on 
>performance - it seems like everyone involved in code 
>generation knows what they are doing though.
> 
>If it is a possiblity I woudl use ANTLR3 now as other than the 
>start up time, while there are potentially bugs in there, they 
>do not matter if you configure your grammar to avoid them and 
>the code you generate works for the task at hand. Future 
>modifications to ANTLR3 don't have to be adoopted until you 
>are ready. Of course, the general patina of Early Access may 
>put many people off but ANTLR3 is close to being 'there' if you ask me.
> 
>Jim
>