Related articles |
---|
Suggestions required please .... markagr@aol.com (1999-04-03) |
Re: Suggestions required please .... dwight@pentasoft.com (1999-04-06) |
Re: Suggestions required please .... dobes@mindless.com (Dobes Vandermeer) (1999-04-06) |
Re: Suggestions required please .... markagr@aol.com (1999-04-09) |
Re: Suggestions required please .... rspartan@usa.net (1999-04-09) |
Re: Suggestions required please .... markagr@aol.com (1999-04-10) |
From: | Dobes Vandermeer <dobes@mindless.com> |
Newsgroups: | comp.compilers |
Date: | 6 Apr 1999 22:47:58 -0400 |
Organization: | None whatsoever |
References: | 99-04-015 |
Keywords: | design |
MarkAGr wrote:
>
> Developers ....
>
> The PDL will be translated into any other language the user
> requires .... (anything from english-prose embedded in HTML to
> assembly language) and so suffers that usual of tri-chotomies
> .... simplicity of understanding vs simplicty of design vs power of
> tool.
Wow.. the language of this message seems to suffer a similar
tri-chotomy.
> Because of the nature of the project's translation technique used
> so far, I would like to be able to compile the PDL into a "semantic
> database / repository" and generate the final language output from
> that. Experience indicates that the language will reflect the internal
> structure of the repository and thus an information model may be the
> best/first port of call, however the repository may be created to
> reflect the language if the language is the right choice. I am willing
> to consider either and other possibilities.
Assuming that what you are looking for is some kind of recommendation,
you could probably get the most flexibility by parsing your top-level
into a relational database that includes as much information (derived
and defined) as possible. This allows the backend to generate code
from whatever information it likes, without being constricted by some
of kind of efficient model.
CU
Dobes
Return to the
comp.compilers page.
Search the
comp.compilers archives again.