This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: EGCS as a component


[warning, quite a bit of flamable material here]

In article <199901190156.RAA19886@atrus.synopsys.com> you write:

>>    There's gotta be a better way to organize the code, to enable tool
>> writers to leverage the remarkably stable front-ends (C/C++ anyway) of 
>> egcs. 

>I'm afraid that's exactly what RMS is extremely worried about:
>specifically, that proposed changes make it easier for *proprietary*
>tools to use the gcc front ends or back ends.

You can view software development from a technical point of view, and
then you can view it from a political point of view.

It is quite understandable that the FSF/RMS might like to push gcc forward
as a nice example of free software development. 

Sometimes, it gets a bit annoying, like the paragraph in gcc.info which
speaks of GNU/Linux, which has absolutely *no* relevance to the documentation.

Sometimes, that means some development problems, like stalling C++,
or not keeping close track on extend.texi, or keeping byte-code in.
We all know that some of those things did lead to the birth of the egcs
project itself.

>The GPL isn't sufficient protection, since you could interface through
>files or sockets or some such to the component.  (The GPL only covers
>code linked into the same executable).

What's nice with the GPL is that people developping software have to
be reasonable... otherwise, someone else will take the ball and be
reasonable with it... 	I could quote Linus, or Eric S. Raymond, but
it is probably unnecessary.
>>   Any other approach taken by egcs would have to enable the rapid
>>   development of the kind of tools listed above.
>
>"would have to"?
>egcs is of course free to completely ignore the problem.

Yep.  My firm belief is that this particular worry of Stallman is completely
political in nature, and does not stand half a minute from a technical
point of view. Being able to have hooks inside gcc is nice. Sooner or later,
someone is going to do it... they can have some commercial purpose, release
the modifed gcc under GPL, and their tools privately... if the resulting
tool is better than base gcc, people will start developping on that base.

The nice point about Open Software development (yep, buzz-word... sorry)
is that it's Open.  Feeling paranoid about some venues of development is
good, but it means finding solutions, not shutting possibilities down.

I would think that, once proper hooks are available in gcc, some 
Open Software developpers will write quite interesting code that takes
advantage of it.   Some commercial developpers will write some interesting
code too. We'll see who wins (I'd wager on Open Software, just because this
is still an open research area). After all, people are still developping 
commercial compilers besides gcc... :)


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]