[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[DotGNU]Re: The Intended usage of GNU software (was Treecc Ruby support)

From: James Michael DuPont
Subject: [DotGNU]Re: The Intended usage of GNU software (was Treecc Ruby support)
Date: Wed, 6 Nov 2002 14:59:47 -0800 (PST)

--- Rhys Weatherley <address@hidden> wrote:
> *sigh* This is getting very monotonous, and getting us nowhere.

Rhys, thank you for giving me an opening here.

> If you want to add multi-language support to treecc, then nothing
> is stopping you.  However, I gave you my considered opinion that
> such a thing is more difficult than you think, and would probably
> end up being a horrible kludge.  That is my opinion, as the author
> of treecc and the world's foremost expert in its design.  Feel free
> to prove me wrong.  With code, not words.

Rhys, that sounds like a difficult challenge. 

The treecc nodes will not be that difficult to wrap and create a
inline/swig interface. That is what I started on. The challenge is to
give it less, not more. If you generate your own accessor methods,
then they get in the way of any language mapping.

Via the swig, or inline, you get the Glue.
The glue is what sticks programs togeather. 
This glue could be XS (from perl), inline specific or even SWIG or some
future Inline::Glue::DotGNU.

> The GNU project may encourage things outside the original intent,
> but it also encourages people to get off their backsides and do
> it themselves!  I have decided that such a feature isn't useful
> for me to look into at this time.  That shouldn't stop you.

great. I will put the treecc interface as a project to interface the
dotgnu project into more languages. My primary intent is to use it as
some form of input (in the form of parsed treecc objects) and output
(as tc files). 

We have three active developers on the introspector, 
and the inline people support the effort. I will be working on the
Swig::Glue modules for the next version of inline. There I will try and
fit the treecc interface in.

> If your patches actually enhance treecc, are consistent with the
> style of the existing code, and don't break anything important, I
> may even incorporate them.  If I won't incorporate them for some
> reason, and you don't like that, then feel free to fork the project.

Up to now, I have not patched treecc, only have ported it, supported
it, tested it, even recommended it.

The only thing that I could expect from the treecc,
is to create an introspector interface for importing data from it.
This could be used then as an input for tree models,and I can then use
the SQL generation for it. 

> If you put half as much effort into code as you do into criticising
> the honest opinions of others, we might acually make some progress.
> We can only break this deadlock if you contribute something other
> than more words.

That sounds like a challenge. I spend alot of time trying to get a
discussion going about some difficult topics. 
My developing time is being spent now on the Win32 port of gnome via
mingw32 under debian. 

We have been busy reactivating the introspector gcc patch. The new
simplified tree structure ssa branch will be the basis for the new
introspector. That will give you a simplifier for each language that
gets rid of complex statements. This gcc interface will also be of
interest to the dotgnu project.

I have done a stich of programming and testing for dotgnu. Yes, I have
done my fair share of bitching. That is why I put all the dotgnu
projects on hold for now. We have to aggree on what needs to be done.

Your the pnet author and project leader Rhys. You decide if you want me
working on the above topics or others or not. 


James Michael DuPont

Do you Yahoo!?
HotJobs - Search new jobs daily now

reply via email to

[Prev in Thread] Current Thread [Next in Thread]