RE: IDE, IDL feedback needed.

["Saber Karmous (DSN)" <Saber Karmous dsn ericsson se>]

> -----Original Message-----
> From:	John R Sheets [SMTP:dusk@smsi-roman.com]
> Sent:	24 June 1998 19:36
> To:	Saber Karmous (DSN)
> Subject:	Re: IDE, IDL feedback needed.
> 
> Saber Karmous (DSN) wrote:
> > 
> > Correct me if I'm wrong, but wouldn't it be a whole lot better if
> you
> > would combine those two, both the line number and the function name
> are
> > important, I think that recent IDE's aren't only using line
> numbering.
> 
> I agree.  We should track the function name too.  This will make
> it easier to provide navigation through the code.  You can say
> "Go to function move_cell()", and not have to worry about
> calculating which line in which module it's on.  The line number
> of the function should be calculated once, up front (and later
> when that file is modified), not every time you want to jump to
> it.
> 
> > You have to track the object hierachy too... I think you have to
> think
> > more radically, IDE's aren't really accessible, the dragging and
> > dropping in all those Visual IDE's is nice but it's limited... 
> 
> Absolutely.  The IDE needs to be aware of the object heirarchy. 
> If you change the name of a member function in C++, the IDE needs
> to respond differently (i.e. make changes in different places,
> using different syntax) than it would to a straight C module, or
> Python, or Scheme, whatever.  This should all be encapsulated.
> 
> Or, for that matter, it's very common to change parameter lists. 
> Each language handles them differently, and (especially with C)
> may allow different syntaxes (which users should be able to
> customize transparently).  
> 
> I think the IDL should represent this, too.  Maybe something like
> this:
> 
> interface Editor {
>   [...]
>   boolean gotoFunction(in string functionName) raises
> (InvalidFunctionName);
>   boolean changeFunctionName(in string oldName, in string
> newName) raises  (InvalidFunctionName);
>  [...]
>  };
> 
> > What would be nice  was a two screen setup, the first screen would
> have you
> > code, the second screen would have a active environment of your
> program,
> > so if you would add functionality it would be active directly on the
> > second screen. The second screen would interpret your code and
> execute
> > it directly... That would be cool... IDE's are nice but the
> 
> Right.  It would be nice to be able to customize that environment
> view, to zoom into the project at whatever depth you needed, e.g.
> global view, object view, file view.  I think the navigation is
> one of the most important features of an IDE.  It can make or
> break the IDE.
> 
> > edit-compile-debug cycle just take too damn long even on a 1000MHZ
> 4GB
> > P13 multiprocessor, it will always take to much time... An IDE
> should be
> > active not passive, the only active part of current IDE's is the
> > debugger... Please consider my thoughts and I hope to hear more
> about
> > this
> 
> What exactly do you mean by an IDE being active/passive?
> 
> John
> 
By active I mean that the application is running in  some kind of
virtual machine (tricky), so that the IDE has an active role when
testing/buidling the app. With passive I 
mean that the IDE only compiles/edits.

Oh yeah I think if this kind of IDE could succeed it would also have a
big impact on how applications are debugged.

Saber

> -- 
>          To unsubscribe: mail gnome-list-request@gnome.org with 
>                        "unsubscribe" as the Subject.