Am 14.07.2006 um 21:20 schrieb Dan Stenning:
I find that a very puzzling thing to say.
Yes, probably.
The goal for using a tool isn't proving the fitness of the tool, but
optimize the efficiency of work and the correctness of results. I
wonder,
if RB is an efficient solution to write compilers.
Since the 'Write RB using RB' is still pending for a very long time now,
doesn't make me feel, that RB is a good solution regarding the above
mentioned
measure. Ever heard of Swordfish? This isn't available now - and
first mentioned
at least a year ago. Heard of CoCoa based RB implementation?
Postponed until
RB migrated to UB.
Please give reasons for why RB might not be suitable for writing a
compiler.
Using a compiler-compiler to automatically generate and compile
scanners,
parsers etc. isn't that simple using RB.
I think RS would be very interested in knowing about such
deficiencies -
they could then implement them.
Due to the continues bug notifications in the ML, the forum and RB's
bug database, RS should have very much insight into buggy parts of
the system.
There should be enough work to do - even if they don't use RB.
What are you suggesting would let down a RB produced compiler?
Since RB is turing complete...
Lack of pointers ? Umm don't think so..
No, not an essential point. At least not big one.
Garbage collection ? Nope - there are many many studies and
papers out
there to show that garbage collected languages can be just as fast
as those
with memory handled "manually" ( C / C++ )
Mostly, speed of an app isn't an issue. Non-handling of cyclic memory
structures
isn't a pro, but most likely a problem.
Lack of templates ?? Nope - a nice to have , but doesn't make for
faster
code.
Not sure.
So why exactly would RB not fit the bill ?
As a customer, I'm mainly interested in a stable and feature rich
solution.
Since most RB users don't write compilers, writing RB in RB isn't a
vital point
for me. I still feel, that the effort binds more time and resources
than other
tools had bound.
E.g. GCC is very long on market, available for many platforms,
maintained
and observed by many people and, thus, probably a good starting point
for
writing a compiler. Mono - although still lacks WindowsForm support on
non-Windows systems, might probably be of interest too. There might
be other
xplat solution [Trolltech] too.
I wrote some compilers using various languages. Personally, I
wouldn't choose
RB for this task. Other may think, that RB is well suited for this
task. No
problem for me.
And, please, don't get me wrong. I don't say that RB is a bad
language in general
here.
_______________________________________________
Unsubscribe or switch delivery mode:
<http://www.realsoftware.com/support/listmanager/>
Search the archives of this list here:
<http://support.realsoftware.com/listarchives/lists.html>
