[antlr-interest] Re: build issues: bytecode assembly generation

[micheal_jor]

--- In antlr-interest at yahoogroups.com, "Don Caton" <dcaton at s...> wrote:
> 
> Terence:

Well, I'm obviously not Ter but...

> I haven't been a subscriber to this list for too long, so forgive me
if this
> has already been discussed, but...
> 
> Why did you write Antlr in Java in the first place, and why are you
fighting
> with these issues?

I suspect because developing in Java is more productive and, Java
delivers cross-OS portability more easily than C++ does. Just think
about the work that would be involved in simply providing binary
builds of the tool [on the website] for all the different platforms
that the Java version runs on today.

> C++ is at least as portable as Java and it has no
> runtime to impose limitations that get the way of what you need to do.

Given that C++ has a goto statement, there would indeed be no need to
generate assembly but the less productive (compared to Java)
development effort over the years would have nullified any advantage
this particular difference represents.

Incidentally, the DFA-as-bytecode generation issue isn't a runtime
limitation, it is a *language* limitation - the lack of a goto [any
label] statement in Java. Ter's intent is to use the runtime to bypass
a limitation in the language.

> I'm
> not anti-Java, but I don't see what benefits it brings to the table over
> C++, at least not for this kind of programming task.  

Better productivity, easier cross-OS portability and a lower barrier
to entry for prospective contributors.

Micheal






 
Yahoo! Groups Links

<*> To visit your group on the web, go to:
    http://groups.yahoo.com/group/antlr-interest/

<*> To unsubscribe from this group, send an email to:
    antlr-interest-unsubscribe at yahoogroups.com

<*> Your use of Yahoo! Groups is subject to:
    http://docs.yahoo.com/info/terms/