Re: [imp-dev] Use of org.eclipse.zest in pdb.ui

[Jurgen Vinju <jurgen@xxxxxxxxx>]

Hi Ed, Bob,

Yes, the visualizations are intended initially for debugging complex
source analyses; although I would
not be that surprised if people start building nice end-programmer
tools from them.

And, in principle I am all for fine-grained and explicit dependencies
between projects and features.
In my experience it increases the chance of use and also the
understandability of the software.

I've no idea about memory footprints of Zest or draw2d;
I am just sailing - happily and obliviously - on the winds of software reuse.

Cheers,

Jurgen

On Wed, Feb 4, 2009 at 6:54 PM, Robert M. Fuhrer <rfuhrer@xxxxxxxxxxxxxx> wrote:
> Actually, as I see it, there's a significant *skilled* user base that
> probably
> wouldn't bother installing GEF, were it not for such a dependency. I
> usually don't install it, for want of any compelling reason.
>
> Also, one good general principle is "the fewer dependencies, the
> better" (all other things being equal).
>
> That said, looking at the GEF download site, the size of Zest + Draw2D
> is around 3M, while for all of GEF is about 8M, so that's arguably not a
> huge difference (compared to the rest of the Eclipse SDK). And you're
> probably right that including all of GEF is more user-friendly than
> requiring just a part of it.
>
> So, IMHO it's fine for pdb.ui to depend on GEF, but I'd still like to put
> the pdb UI code in a separate feature with the appropriate feature-
> level dependencies.
>
> Remember, the default/generic visualizations in pdb.ui will probably
> be more useful for development and debugging of IDEs than for the
> end-users of the IDEs.
>
> On Feb 4, 2009, at 12:04 PM, Ed Willink wrote:
>
>> Hi
>>
>> Is there a significant unskilled user base that needs to not have GEF?
>>
>> There are numerous projects that depend on GEF so surely everyone has it.
>>
>> Requiring only half of GEF is more confusing than requiring all of it.
>>
>>        Regards
>>
>>                Ed Willink
>>
>>> -----Original Message-----
>>> From: imp-dev-bounces@xxxxxxxxxxx
>>> [mailto:imp-dev-bounces@xxxxxxxxxxx] On Behalf Of Robert M. Fuhrer
>>> Sent: 04 February 2009 14:12
>>> To: IMP Developers List
>>> Subject: Re: [imp-dev] Use of org.eclipse.zest in pdb.ui
>>>
>>>
>>> Hi Jurgen,
>>>
>>> No question that a zest dependency is superior to the
>>> cmd-line graphviz
>>> dependency. Somehow I was under the impression that draw2d was
>>> part of SWT itself (just naive hope, I guess).
>>>
>>> I don't think it makes sense to extract the
>>> graph-visualization code to
>>> another plugin, since it's the main bit of functionality in
>>> the plugin
>>> as
>>> it is (as you said).
>>>
>>> At the same time, it may make sense to extract pdb.ui to another
>>> feature,
>>> so that people can have the core analysis representation and
>>> computation code without dragging in GEF, if they have no need for
>>> visualization beyond tables and lists.
>>>
>>> On Feb 4, 2009, at 2:57 AM, Jurgen Vinju wrote:
>>>
>>>> Hi Bob,
>>>>
>>>> Well, Zest depends on draw2d which is a part of GEF, and the
>>>> DirectedGraphLayout algorithm, which is also used by GraphViz' dot
>>>> tool, is a part of draw2d.
>>>> I consider this a superior alternative to having a dependency on
>>>> graphviz commandline tools, or to implementing the layout algorithm
>>>> again ourselves.
>>>> I have also understood from the authors of Zest that pieces
>>>
>>> of it are
>>>>
>>>> sometimes moved to draw2d if deemed necessary.
>>>>
>>>> We could factor uit the graph dependent code to a new project called
>>>> pdb.ui.graph. On the other hand, I'm not sure that would
>>>> help since the graph drawing is what is most useful anyway.
>>>>
>>>> Cheers,
>>>>
>>>> Jurgen
>>>>
>>>> On Wed, Feb 4, 2009 at 1:13 AM, Robert M. Fuhrer
>>>
>>> <rfuhrer@xxxxxxxxxxxxxx
>>>>>
>>>>> wrote:
>>>>> Hi Jurgen,
>>>>>
>>>>> Is Zest part of GEF, or can it be obtained separately? I'd
>>>
>>> like to
>>>>>
>>>>> avoid
>>>>> making pdb.ui dependent on something as large as GEF, but not,
>>>>> of course, if there's no good alternative...
>
> --
> Cheers,
>  - Bob
> -------------------------------------------------
> Robert M. Fuhrer
> Research Staff Member
> Programming Technologies Dept.
> IBM T.J. Watson Research Center
>
> IMP Project Lead (http://www.eclipse.org/imp)
> X10: Productivity for High-Performance Parallel Programming
> (http://x10.sf.net)
>
> _______________________________________________
> imp-dev mailing list
> imp-dev@xxxxxxxxxxx
> https://dev.eclipse.org/mailman/listinfo/imp-dev
>