Re: Prothon is switching to the .NET platform

On Thu, 05 Aug 2004 22:12:45 -0700, Guido van Rossum wrote:

> Hi Mark,
> 
> It was a pleasure to meet you at VanPy, and I was sorry that I had to
> leave in the middle of your talk (I had a plane to catch and David had
> scared me about delays).
> 
> I was pleasantly surprised in hearing your architecture for the
> Prothon engine -- it sounded like a fresh look at implementing a
> runtime for a Python-like language, and several features (like the
> locking strategy) sounded like I should be paying attention to them
> for the Python 3000 interpreter.

Thank you.  That means a lot to me.

> So now I'm a bit disappointed to hear that you're giving all that up
> and switching to the CLR.  I understand your desire to have a large
> library available (though most of that library will be class-based,
> which may require you to bring classes back in through the back door).
> But in practical terms, it makes Prothon a lot less accessible in the
> years to come.

Both Dave and Paul pointed out IronPython to me at the workshop and how it 
could get Prothon a library and therefore a community much faster than 
going it the hard way like you did.  There is a general feeling going 
around that multiple language VM's are the "in-thing".

I've got a kind of catch-22 that I'm worried about.  I have no one helping 
me write libraries because they aren't using Prothon and no one will use 
Prothon without libraries.  They got me thinking that .Net would break the 
catch-22.

Do you have any other ideas on how I could get a decent library in a decent 
timeframe?

> I recall that installing .NET on Windows took me half a day.  Maybe
> it's gotten better, but I sure can't believe they've shrunk it.  Mono
> gives you theoretical support on Linux, but it's even more of a
> questionable proposition to actually run it (and its set of supported
> runtime libraries has only a small intersection with what's available
> in .NET).  Basically, telling people "here's my language, it's a 2 MB
> download, and oh, BTW, you also need to install the 200 MB runtime
> over there" isn't exactly encouraging people to play with your
> language.  Even for apps written in Java this is a significant barrier
> -- I have several copies of Java on my hard drive because every
> significant application written in Java ends up needing a different
> JVM version.  And the JVM I downloaded last year is probably too old
> to run this year's crop of bleeding edge Java apps.  Availability of
> the CLR is a lot less than Java.

.Net and Mono may not be the greatest solution in the world, but they have 
a shot at running Prothon.  I looked into Parrot and CPython and neither 
one could do the job. 

> I'm also surprised because you admit to being a poor language designer
> (although you picked a great example :-).  I assumed this to mean that
> you're having more fun implementing the runtime.  But if you're not
> doing it for the language design, and now you're giving up on the
> runtime, what's left?

Just because I am not a good language designer does not mean I am not a 
good project manager and programmer.  I would not enjoy working on the fun 
parts of the VM knowing that some other path existed where the language and 
community could grow faster.  I guess I'm finding out I'm more motivated by 
Prothon's overall success than "programming fun".

There will still be a lot of fun left.  IronProthon will still be prototype 
based.  It's a long story but the CLR will be capable of running my 
prototypes, the co-routines, and everything in Prothon except light-weight 
threads, object locking, and the locking-based features.

> Hoping your website was hacked or it was practical joke

The anti-.NET forces are just starting to appear now on the Prothon mailing 
list.  It is starting to get interesting.  I can always change my mind as I 
have many times in this project.  I'm kind of famous for that.  Right now 
though the idea of having Prothon be something that could be truly useful 
in a matter of months instead of years is very attractive to me.

BTW: I hope my new website verbage isn't offensive to you.  In the past I 
wasn't considering Prothon a successor to Python as I was leaving that for 
Python 3000, but when you said that 3000 was mostly going to change range 
to an iter I had a change of heart. :-)  Were you serious about that?

Mark Hahn