discuss-gnustep
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Project Center inspiration sshots


From: stefan
Subject: Re: Project Center inspiration sshots
Date: Mon, 3 Jan 2005 16:35:45 +0100
User-agent: Internet Messaging Program (IMP) 3.2.2

Citát Björn Giesler <bjoern@giesler.de>:

> Hi,
> 
> Am 03.01.2005 um 13:43 schrieb Philippe C.D. Robert:

<snip>

> > As the name implies, PC should IMHO be the tool which glues together 
> > all steps of a development project. Hence PC shouldn't become a 
> > specialised tool itself, but it should seamlessly integrate such tools 
> > into the development process if needed! Right now there is just Gorm 
> > as an example, but other such tools could be a good visual debugger, a 
> > profiler, a class browser or an UML tool of course.
> 
> Yes, I totally agree. I don't suggest this for PC proper, although 
> Stefan did with his screenshots. I think it's a workable paradigm for a 
> development component ("editor"?), though, which could be available 
> from ProjectCenter. Is there an interface definition for external tools 
> to interact with ProjectCenter? If not, I think there should be :-)
>

I am not sure that I can say that I agree. PC despite of its name is (should
be)
an IDE. IDE means Integrated Development Environment. User should feel that the
project parts are integrated, not scattered around. If they do not get the
feeling with PC, they will use another IDE. 

But back to the point. PC should not be just a file-integrating tool with build
abilities. Then what is the difference between combination of a file manager +
editor (on doubleclicking the file) + template shell scripts for
build/run/debug (doubleclicking executables in a project directory)? Yes, it is
exagerrated for better illustration, but you see that there is no difference in
functionality.

As of UML: yes, it can be useful. But at this moment it is too far to be
achieved. I would like to have some graphical code modeller, but be more
realistic. Reason why I was suggesting several links for screen shots and
already existing code was that the refered things were immediately available. 

To be more concrete here is an example. It is easier to build smalltalk-like
browser than to build UML editor. For the browser you roughly need:
- NSBrowser for classes (**)
- NSBrowser for methods (**)
- text view for method code (*)
- view for editing class properties + additional info (such as required
non-standard include files)
- model classes: "project", "class", "method", "ivar" (*)

(*) - already implemented
(**) - joined by some smalltalk browser controller

For the UML you will need many many new classes that are not yet implemented,
such as diagram classes (diagram, diagram item, connection) or many graphics
related methods.

It is better to stay on the ground, but think about the possible enhancements.
That means, that by doing first version of PC we should think that there can be
UML (or any other modelling) editor, but we should not focus on it right now.

Conclusion:

- focus on human thinking while SW development and aid the thinking process
instead of focusing on currently available development
tools/processes/elements
- split goals to immetiately doable goals and future goals
- PC should be IDE and should feel like IDE

Regards,

Stefan Urbanek



reply via email to

[Prev in Thread] Current Thread [Next in Thread]