Re: About o:XML

On Tue, Apr 19, 2005 at 10:41:27PM +0000, Brian Chrisman wrote:
> I'm in pretty much violent agreement with all your points.

> Choosing a language for which it is relatively easy to write
> a compiler allows you greater choice in whose compiler you
> will use and what features will be available.  I think if I
> were to write a non-XML-syntax language right now, I'd pre-parse
> it into an XML document anyways.... Brainfsck may be a logical
> extrema in this argument, but we're generally talking about
> useful languages, I think.. :)

Thinking one step further here will probably lead you towards compilers with a
plugin architecture. Currently we can only choose between compilers from
different vendors, but assuming an easily parseable XML as the base there's a
wide range of possible improvements that could easily be incorporated into a
compiler framework through plugins.

> btw, have you (or for that matter, anybody) heard of any source management
> systems that are tailored for XML-language development?
> In particular, I thought it would be pretty nice to have such a system
> that will report context of changes made... some such systems do this
> for other languages...

No.. and yes :)
I haven't heard of an existing one yet, but the SrcML project is currently
a university project here with several students developing applications on top
of the SrcML representation (and the provided API). One of those projects is a
source code management system which takes advantage of storing the code in
XML. But I suggest you don't hold your breath, as it will at least take a
semester before it's useable and I don't know whether it will actually reach a
production state at all.

However I'd appreciate if you could get into more details about what you
expect from such a system. What kind of queries would you want it to be able
to answer for example, where do you see the major advantages to line-based
SCMs, etc. I suggest you answer to me directly or through the srcml-discussion
list, as this isn't related to o:XML now anymore.

-- 
Raiser, Frank
Student @ University of Ulm (www.uni-ulm.de)

Get and set methods are evil. (Allen Holub)