[Prev][Next][Index][Thread]
Re: MIME for Dylan sourcecode?
"Olivier Lefevre" <Olivier.Lefevre@genedata.com> wrote in message
3b9bac36@news.core.genedata.com">news:3b9bac36@news.core.genedata.com...
> > One could imagine an IDE that used a database derived by slurping in
text
> > files and spitting them back out in some well-defined and reproducible
> > form for interaction with a revision control system.
>
> Isn't this how IBM's VisualAge works? It is not file-based, either, and so
> faces the same issues.
FWIW, FunO's Dylan IDE slurps in the files and creates an
in-memory "database" of each file's definitions. The compiler
adds additional informatiom to this when you compile the
projects. The compiler database remembers much of this
information, but the "sectionization" of files into definitions is
done when you read the file; there's no point in saving it.
You can see what FD's IDE does with this by, for example,
clicking right on a class name in the editor and selecting
"Edit Methods" from the pop-up menu: it creates a "composite"
buffer consisting of all the methods for the class. This buffer
can be saved out, and the definitions are all written to their
proper home files.
And yeah, I know it's not how Smalltalk IDEs work. I happen
not to like how Smalltalk IDEs work, and decided to try it a
different way.
Follow-Ups:
References: