Re: [orbitcpp-list] announcing o2cpp

[Phil d. <ph...@us...>]

Andreas Kloeckner wrote:
> 
[...]
> > > * Choice of language
> >
> > Fair enough (BTW, it's only the compiler that's written in python - all
> > of the runtime is and will be in C++). What didn't you like about
> > python?
> 
> True, it is only the compiler. But all this
> get-python-install-automake-stuff-get-everything-to-work-retry-several-times
> can really suck, and I believe it can scare people off, even coders.

True - from an installation point of view, the python build stuff is a
bit of a pain if you don't know python. (The next official version of
automake (1.5) will have python support, but that's beside the point)

> People that are looking for a c++ binding do know c++, but they might
> not know python. If they wanted to help properly, they'd have to know
> both.

Again true. I've been convinced that ideally from a community
perspective the compiler should be in C++; it's just that I didn't
really want to write it in C++ myself (too much like hard work!) - I
wasn't sure about the compiler design, and it's much easier for me to
refactor python code than C++ code (no declaration redundancy ;-). 

> (You want to know what i disliked about python? oh, plenty. not that i
> had actually tried, but there were several things that pissed me off
> enormously. first, python is not as declarative, as, say, c++. this
> makes code generally a bit harder to read. also, python does not try to
> separate declaration and implementation. 

But typically, due to python's terseness, your python class (inc
implementation) takes up less space than the header file in C++ anyway.
I'd speculate that I could grok a number of python classes in a single
file quicker than the same code spread over a number of C++ class files.
(don't even get me started on Java and its one-file-per-public-class
approach - eugh!)

> second, qualifying everything
> with "self." does not contribute to either code legibility or coding
> speed. Of course, the need for extra qualification is a direct
> consequence of having no declarations. 

Yep - you're not the first to dislike the self thing. As you say, the
self qualification is essential because you must differentiate between
object instance scope and method scope, and you don't have declarations
to do that.
On the other hand, if you're just talking about coding speed,
declarations do add redundancy. It's a static vs. dynamic typing thing -
static declarations are more work but safer. Decide what is important
and pick the right tool for the job. 

> third, i did at first not like
> the idea of a language being picky about whitespace. (I have changed my
> opinion a little on this one) ... <insert more senseless criticism
> here>... )
> 

Fair enough. I've never found it a problem personally.

> > What's your plan? Do you want to join forces / merge code?
> 
> Yup. :) My proposal would be to base the "new thing" (tm) upon your new
> runtime code (see other mail) along with an adapted version of my idl
> compiler. Which sucks because it is loads of work to do. Not that I'd
> mind, but currently I'm facing one and a half month featuring five
> exams, in two weeks' distance each. I guess I won't be able to do much
> of the like until March 24, when the exams will finally be over. After
> that date, however, I've got six weeks of free time (being a student is
> not so bad after all :).

Yep, I remember that... (sigh)
I wouldn't swap student life for working life though - all that time
sitting around in bars feeling guilty because I should be working ;-)

> We might start real soon by putting everything into one cvs tree,
> however.
> 

I'm currently doing a prototype (proof-of-concept) with the current
ORBit-C++ runtime and the string test. I wanted prove that it would work
before committing anything to CVS. 

Would it be okay from your standpoint to add your compiler to the
ORBit-C++ source tree and then refactor the runtime from that?

BTW, Do you have gnome cvs access?

> >
> > (BTW, I've put my thoughts and comments about the o2cpp implementation
> > in another mail)
> 
> (your fault. you'll have to read two replies ;)
> 

On the contrary - the more time I spend writing emails, the less I spend
doing 
'real-work' (tm)

Cheers,

Phil 'hoping his employer doesn't read orbitcpp-list' Dawes