netron-graphlib — The Netron Project Mailing List

[Netron-graphlib] Bezier connections and a little inquiry

[Francois V. <Fra...@sk...>]

(Posted in the forum as well)


While looking at intermediate connection points (it's finished by the way) I
was wondering how to implement curved connections. It's actually an idea I
had a long time ago; the GDI library allows you to draw Bezier curves but
there is no parametrization. For example, what are the coordinates of the
intermediate points if you pass a few (PointF) points to the
Graphics.DrawBezier method? Annoying thing, so I developed my own version.
You can find it here: http://netron.sf.net/downloads/BezierControl.zip
It's not integrated in the graph library, though it's now a piece of cake to
do it. If you look at the code of the Bezier control you 'll notice that it
contains the essential of the graph control: how to use the mouse events to
control GDI objects.

Concerning the difficulties with the cursors, I shifted all cursors to the
upper-left corner and I hope this helps. Actually, are the cursors OK for
you? 

I'll look at the other issue/bug you raised tomorrow (Martin). 


Now, a little story.
Had some feedback from some co-workers in the office who discovered I'm the
owner of Netron (argh, don't like to be discovered). Surprising thing is
that to many people it's not obvious what the graph library is and if they
know they find it too difficult to understand. Kinda shocking to me. I
really try to make it accessible by means of various papers and examples, a
lot of Wiki pages, forum, news list etc....seems still too abstract for
some. Well, maybe I've to conclude that it says more about those people than
about the library. Not sure though. From someone in the forum I understood
that the architecture paper is a blessing and I should finish or add
tutorials...
So, my question to you (who's leastening, I don't know): is the graph
library complicated? 

I sometimes have the impression that by releasing the source code people get
confused. If you buy a control and you get a CD with example and
documentation but no code things seem more transparent. In trying to explain
things in depth some (most) get lost, no?

Keep well, Francois.

Re: [Netron-graphlib] Black-box shapes

[Martin C. <mar...@ei...>]

Hmmm, interesting. I have to do something similar. I needed to be able 
to collapse all child nodes into one shape, and then expand them again 
if asked. This may be done at multiple levels in the diagram. Besides 
the connectors and connections issues raised by Francois, there is also 
the issue of layout. When you box a group of shapes, you end up with one 
shape in place of N shapes and lots of free space.
Joe User then decides to add new shapes in the newly created 'free 
space'. Afterwards he decides to unbox the previous group of shapes, 
where do they go? Do we put them back in their original unboxed positions?

Regards,

Martin

Illumineo wrote:
> For JR.
> 
> On a few occasions people have asked something like 'group shapes', a kind
> of box/unbox feature. I understand the need for this very well; if you have
> a lot of shapes things start to become ugly and difficult to overview.
> Besides this, moving related shapes becomes more easy in this way. In
> certain applications this is called a 'Xshape' and can be saved separately
> (I think of Cinema4D here). So...yes, I've thought about it but it never got
> into anything concrete. I think it's not the kinda thing you can do before
> the rest of the code is stable...which is the case now (tell me if it's
> not!). Saying this, it's not a simple thing to implement. How to handle the
> connections and the connector? I can see some straightforward code for
> grouping shapes but I cannot immedaitely see how to handle the links and
> connectors.
> 
> If anybody has some code or ideas, gimme some feedback.
> 
> For Martin.
> 
> How are things going with GraphML? I haven't had time this weekend to do
> enything Netron-like but I woke up at 5AM this morning to dive into it
> again. Don't ask me how I feel now ;-)
> 
> All the best, I.
> 
> 
> 
> 
> 
> 
> ____________________________
> Hi,
> The new v.2 is working great, thanks for all the hard work.  
> 
> I'm interested in creating a shape that encapsulates a set of other shapes
> within it.  Once some functionallity is determined and tested for a set of
> shapes I want those shapes to be able to be hidden from the user's view in
> the graph control.  This way a single so called "Blackbox Shape" that hides
> the internals from the user will be displayed in their place.
> 
> My initial plan is to define a shape with its own GraphAbstract in which I
> copy the set of shapes into and remove them from the GraphControl's extract
> or simply load them from a saved file into the Blackbox shape. With some
> user input the Blackbox's Abstract would be displayed either in another
> form's GraphControl or something...not quite sure until I try it.
> 
> I was wondering if you had experienced any issues with this or have tried
> something simialiar to this during your development of the NetronGraphLib
> and the sample apps?
> 
> Thanks,
> -JR
> 
> 
> -------------------------------------------------------
> This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
> Use IT products in your business? Tell us what you think of them. Give us
> Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
> http://productguide.itmanagersjournal.com/guidepromo.tmpl
> _______________________________________________
> Netron-graphlib mailing list
> Net...@li...
> https://lists.sourceforge.net/lists/listinfo/netron-graphlib
> 
> 

Re: [Netron-graphlib] Subpoints on connections

[Martin C. <mar...@ei...>]

Hi Francois,

As you have fixed the bugs, can you let me know which ones you've fixed 
via email or IM? I'll probably have a look at them tomorrow, so I don't 
want to be duplicating the effort!

Subpoint on connection half done, great. I began looking at it this 
evening to see what might be involved. With regard to sub-contracting 
the best I can do for now is to contribute back to the project ( it is 
open source after all)  :-) . Here's what I hope to have for next week, 
some of which is already done:

1. User defined connection for GraphControl.
2. New connection style -  Curved.
3. GraphML support.

Regards,

Martin

Thanks for the bugs Martin, I'll have a look.

The subpoints on a connection is half done, it'll take another few days to
iron out some things. You should start thinking about sub-contracting
me...;-)

GraphML is indeed not a pudding and I fear that every new little addition
breaks the XML schema. That's why I kept pushing it behind.

Keep well, I.




-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl
_______________________________________________
Netron-graphlib mailing list
Net...@li...
https://lists.sourceforge.net/lists/listinfo/netron-graphlib

[Netron-graphlib] Subpoints on connections

[Francois V. <Fra...@sk...>]

Thanks for the bugs Martin, I'll have a look.

The subpoints on a connection is half done, it'll take another few days to
iron out some things. You should start thinking about sub-contracting
me...;-)

GraphML is indeed not a pudding and I fear that every new little addition
breaks the XML schema. That's why I kept pushing it behind.

Keep well, I.

[Netron-graphlib] Bugs, GraphML and request

[Martin C. <mar...@ei...>]

Hi Francois,

I've come across a few bugs in the current code. I'll list them here so you
have a record. I may fix them myself time permitting.

1. Using the scrollbars to scroll to the left or to scroll up on a big graph
leaves lines on the graphcontrol. This lines disappear when you move the
mouse over the graphcontrol
2. When using CTL+SHIFT+Right mouse button to Zoom Out the popup menu for
the control pops up.
3. Dragging a connection from one Connector to another can be a bit tricky
sometimes. Grabbing the from connector is fine but getting the hit on the to
Connector is problematic. I think the root of the problem is that the
crosshair cursor is offset from the line end and therefore a hit event is
not raised when the crosshair moves over the connector. Maybe the problem
can be solved by changing the cursor. (This was picked up by people using
this for the first time here)
4. Trying to resize a shape that has connectors defined has been causing me
problems too.

GraphML support - argghhh ;-) Tricky thing to handle. I've got some basic
stuff put together, but its not generic enough. I hope to have something
done for next week Wed/Thurs.	

I now have a need for adding connectors to connections. Is this something
you would consider adding as standard? 

Regards,

Martin

[Netron-graphlib] Black-box shapes

[Illumineo <Net...@ho...>]

For JR.

On a few occasions people have asked something like 'group shapes', a kind
of box/unbox feature. I understand the need for this very well; if you have
a lot of shapes things start to become ugly and difficult to overview.
Besides this, moving related shapes becomes more easy in this way. In
certain applications this is called a 'Xshape' and can be saved separately
(I think of Cinema4D here). So...yes, I've thought about it but it never got
into anything concrete. I think it's not the kinda thing you can do before
the rest of the code is stable...which is the case now (tell me if it's
not!). Saying this, it's not a simple thing to implement. How to handle the
connections and the connector? I can see some straightforward code for
grouping shapes but I cannot immedaitely see how to handle the links and
connectors.

If anybody has some code or ideas, gimme some feedback.

For Martin.

How are things going with GraphML? I haven't had time this weekend to do
enything Netron-like but I woke up at 5AM this morning to dive into it
again. Don't ask me how I feel now ;-)

All the best, I.






____________________________
Hi,
The new v.2 is working great, thanks for all the hard work.  

I'm interested in creating a shape that encapsulates a set of other shapes
within it.  Once some functionallity is determined and tested for a set of
shapes I want those shapes to be able to be hidden from the user's view in
the graph control.  This way a single so called "Blackbox Shape" that hides
the internals from the user will be displayed in their place.

My initial plan is to define a shape with its own GraphAbstract in which I
copy the set of shapes into and remove them from the GraphControl's extract
or simply load them from a saved file into the Blackbox shape. With some
user input the Blackbox's Abstract would be displayed either in another
form's GraphControl or something...not quite sure until I try it.

I was wondering if you had experienced any issues with this or have tried
something simialiar to this during your development of the NetronGraphLib
and the sample apps?

Thanks,
-JR

[Netron-graphlib] UDC's

[Martin C. <mar...@ei...>]

Francois,

I have gotten the UDC to work as follows:

Assembly[] arOfAssemblies = AppDomain.CurrentDomain.GetAssemblies();
ObjectHandle obj = null;
foreach( Assembly _asm in arOfAssemblies )
{
     if( _asm.FullName.Substring( 0, 14 ) == "WorkflowShapes" )
     {
         Type[] arOfTypes = _asm.GetTypes();
         foreach( Type _type in arOfTypes )
         {
             if( _type.Name == "Rule" )
             {
                 obj = AppDomain.CurrentDomain.CreateInstance
                         (_asm.FullName,
                         _type.FullName );
                 break;
             }
         }
     }
}
connection = (Connection)obj.Unwrap();


I need to refactor it to remove the hard coding and to use properties, 
but this code loads the Connection class by name. Now considering the 
fact that the shape libraries can be loaded at runtime, I'm thinking 
this code should execute after any call to LoadLibraries(). What do you 
think?

BTW: I'm going to have a go at GraphML support over the next few days. 
Are you  available to bounce ideas off? :-)

Regards,

Martin

I have no time before tomorrow to look closer at the code you wrote but I
can tell you the kind of problem I often encountered in the past.

Assume you have two identical classes, one inside your main application and
one put in a separate assembly. Next you reflect the latter one in your main
application...it's *not* the same class as the one in your main application
because the CLR takes also the signature of the assembly it belongs to. For
example, on trying things like

If typeof(myReflectedInstance)==Blabla.MyClass ...

This will return false! The way out it using interfaces, if you assign a
reflected class to an interface it will work. Netron uses this a lot, have a
look at the graph lib reflection or the way NAF imports plugins. The
disadvantage of this is that everything you want to use in your main app
needs to be defined (via interfaces). Well, not really a problem, seems a
good thing to have things under control when reflecting. I believe the
divergence between classes and reflected identical classes is related to the
way .Net is imposing security. There is a lot of stuff going on behind the
screen with respect to security, even if one doesn't use it.

Hope this helps you, if you cannot find a way out I'll look at it in the
weekend.

By the way, if you use Messenger you can reach me at
'NetronProject(@)hotmail.com'.

See you!




-------------------------------------------------------
This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
Use IT products in your business? Tell us what you think of them. Give us
Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
http://productguide.itmanagersjournal.com/guidepromo.tmpl
_______________________________________________
Netron-graphlib mailing list
Net...@li...
https://lists.sourceforge.net/lists/listinfo/netron-graphlib

[Netron-graphlib] UDC's

[Francois V. <Fra...@sk...>]

Martin,

I have no time before tomorrow to look closer at the code you wrote but I
can tell you the kind of problem I often encountered in the past. 

Assume you have two identical classes, one inside your main application and
one put in a separate assembly. Next you reflect the latter one in your main
application...it's *not* the same class as the one in your main application
because the CLR takes also the signature of the assembly it belongs to. For
example, on trying things like

If typeof(myReflectedInstance)==Blabla.MyClass ...

This will return false! The way out it using interfaces, if you assign a
reflected class to an interface it will work. Netron uses this a lot, have a
look at the graph lib reflection or the way NAF imports plugins. The
disadvantage of this is that everything you want to use in your main app
needs to be defined (via interfaces). Well, not really a problem, seems a
good thing to have things under control when reflecting. I believe the
divergence between classes and reflected identical classes is related to the
way .Net is imposing security. There is a lot of stuff going on behind the
screen with respect to security, even if one doesn't use it.

Hope this helps you, if you cannot find a way out I'll look at it in the
weekend.

By the way, if you use Messenger you can reach me at
'NetronProject(@)hotmail.com'.

See you!

[Netron-graphlib] Additions

[Francois V. <Fra...@sk...>]

Good morning,

You can find in the Netron (hidden) downloads:

1. http://netron.sf.net/downloads/ForMartin.zip It contains the
custom-context-shape-dependent-menu and the downward mouse-events delegates
to the Entity. Start the Generic graph application up and ckick on the first
example (in the 'examples' menu item). The text-label has the 'special menu'
in the context. The menu is attached to shapes, not to instances though this
is possible as well I guess. The 'Node1' has a mouse-up handler attached and
shows you a message box on releasing the mouse. You can try with the
mouse-down event, it works as well (commented it out in the code). There is
the 'docs' directory a first draft of my technical paper, can you have a
look at it and tell me if it covers everything, if it needs something I
overlooked? It'll take a month or more to finish it but it'll be valuable I
think. 

2.  http://netron.sf.net/downloads/SimpleVB.zip This is a VB example
delivered by Helge who was going on a trip for a few days and did not want
to decieve me ;-) Well, it needs some polishing but it shows clearly that VB
is not a problem. 



For Jason! Could you send me an example of your ASP.Net application? I would
like to include it in the release. 




Have a nice day all, the autumn color are wonderful overhere. It compensates
the depressing Belgian lead-sky :-)

Re: [Netron-graphlib] User defined connections

[Martin C. <mar...@ei...>]

Hi Francois,

Sorry to hear about your mother, I lost my Grandmother to cancer a few 
years ago, I know how hard it can be.

Maybe user defined connections is the wrong term. I don't want to let 
the user drag and drop there own connections but rather I would like to 
be able to tell the GraphControl to use a Connection that I have 
defined, in place of the standard one. For example I have used a spline 
curve connection in the past. I also want to add additional properties 
to the connection for the project that I'm working on. These kinds of 
things would apply to anyone trying to build an application on top of 
Netron. A developer of a UML app, would need additional properties for 
the multiplicity.

When can I have the shape dependant context menus and user defined 
actions? :-)

I too have looked at the Java libraries for guidance, there is a lot of 
good stuff implemented here. When I get time I think I'll try porting 
some of these.

Regards,

Martin

Illumineo wrote:
>  
>  Martin,
> 
> It's only now that I see your mail of the 26th, sorry for this late reply.
> My mother is dying of cancer and it absorbs me now and then.
> 
> The user-defined-connections (UDC): the only thing I have in my hands not is
> the stuff you can find in v1b and which you have been modelling since then
> (I guess). So, no there is not anything ready. I have to ask though how you
> see these UDC's because in v2 there are two types of connections: standard
> and rectangular, both with all the other connection attributes (arrow,
> colors...). Is this a UDC? I guess you see a UDC as an inherited connections
> class but in this case I understand that this is part of a Visio-like idea,
> which I'm not frantic about as you know. This is 'cause it does not support
> any data flow, a freely moving UDC does not enforce connector-based
> connections. The easiest way to implement a freely moving UDC is by using
> the Shape class which would give a thing that 'looks' like a connection but
> is really just a shape and, hence, not forced to belong to any connectors.
> 
> For the other two items: I think I can implement shape-dependent
> context-menus is a snap, as well as the delegates for user-defined actions
> on the Entity level.
> 
> I like the 'family' idea of layout algorithms. It is something I already
> tried if you look at older downloads (circular algorithm, and an algorithm I
> found in DrDobb's). Unfortunately they gave hyperbolic results. The easy way
> is to look at the Java libraries out there, many many good graph libs in
> Java. In that respect, we still have to go catch a lot of stuff Java
> developers already did.
> 
> Best regards, Francois
> 
> 
> -------------------------------------------------------
> This SF.Net email is sponsored by: YOU BE THE JUDGE. Be one of 170
> Project Admins to receive an Apple iPod Mini FREE for your judgement on
> who ports your project to Linux PPC the best. Sponsored by IBM.
> Deadline: Sept. 24. Go here: http://sf.net/ppc_contest.php
> _______________________________________________
> Netron-graphlib mailing list
> Net...@li...
> https://lists.sourceforge.net/lists/listinfo/netron-graphlib
> 
> 

[Netron-graphlib] Network loaded images labels (Dan Gould)

[Francois V. <Fra...@sk...>]

I see your point Dan. It's the kinda thing you usually see in mind-mapping
software (Mindjet.com). The image-thing is really easy to do, actually there
is an example in the v1b version (an image shape).

The random placement of nodes: this is done is many places. See for example
the way the graph-analysis is implemented, the generated random graphs are
both random in structure and in coordinates.

The code is something like:


for(int k=0; k<nodeCount;k++)
{
	shape=
this.graphControl1.AddShape("17110CC6-CCA2-41a1-8963-86CEBA753266",new
PointF(10,10));
	shape.Text = k.ToString();
	shape.FitSize();
	shape.X=rnd.Next(30,this.graphControl1.Width-30); //keep it a little
off the edge of the canvas
	shape.Y=rnd.Next(30,this.graphControl1.Height-30);
	matrix.AddVertex(k);
	nodes[k] = shape;
}

Kind regards, Francois.

[Netron-graphlib] User defined connections

[Illumineo <Net...@ho...>]

 
 Martin,

It's only now that I see your mail of the 26th, sorry for this late reply.
My mother is dying of cancer and it absorbs me now and then.

The user-defined-connections (UDC): the only thing I have in my hands not is
the stuff you can find in v1b and which you have been modelling since then
(I guess). So, no there is not anything ready. I have to ask though how you
see these UDC's because in v2 there are two types of connections: standard
and rectangular, both with all the other connection attributes (arrow,
colors...). Is this a UDC? I guess you see a UDC as an inherited connections
class but in this case I understand that this is part of a Visio-like idea,
which I'm not frantic about as you know. This is 'cause it does not support
any data flow, a freely moving UDC does not enforce connector-based
connections. The easiest way to implement a freely moving UDC is by using
the Shape class which would give a thing that 'looks' like a connection but
is really just a shape and, hence, not forced to belong to any connectors.

For the other two items: I think I can implement shape-dependent
context-menus is a snap, as well as the delegates for user-defined actions
on the Entity level.

I like the 'family' idea of layout algorithms. It is something I already
tried if you look at older downloads (circular algorithm, and an algorithm I
found in DrDobb's). Unfortunately they gave hyperbolic results. The easy way
is to look at the Java libraries out there, many many good graph libs in
Java. In that respect, we still have to go catch a lot of stuff Java
developers already did.

Best regards, Francois

[Netron-graphlib] Feature req: Network loaded images labels

[Dan G. <dl...@cs...>]

First, thank you all for Netron Graph Lib.  I haven't had much of a chance
to play with it yet, but it looks quite nice.

I've got one little feature request.  I suspect I won't get a chance to
code it ymself any time soon, but if no one implements it first and I do,
I'd be glad to contribute the code back.

It looks like lots of the applications are good for showing relations
between data over the 'Net (Groogler, etc).  It would be very cool if you
could set an initial text label for a node but then give it a URL for an
image to replace that label with.

So, at first, it would display the text label.  It would then open an
(async) http connection to grab the image.  Once the image was loaded it
would replace the label with that image.

(Also, it would be nice if you could specify that the initial placement of
nodes on the canvas is random).

Thanks & best,
Dan

[Netron-graphlib] Re: Weekend

[Martin C. <mar...@ei...>]

Good Morning Francois,

Okay so we chuck out MagicLibrary, fine by me. Send over a ZIP file when 
your done and I'll incorporate it into CVS :-)

I'll leave the module aliasing alone for the moment. Let's get things 
stable first.

Source code control systems aside, intentionally checking broken code 
into a repository is bad form. We've all done it in error at some point, 
i.e. we thought the code worked. I think Visual Studios integration with 
   SourceSafe can sometime force you into thinking, I must checkin in 
case I loose something. Also SourceSafe doesn't handle conflicts or 
multiple people working on the same file. I use both source code control 
systems on a daily basis, SourceSafe in work for our own code, CVS at 
home or when dealing with open source projects. Hell I've even started 
to tinker with subversion. I have no problem managing the repository in 
my spare time, but I would be great if you could get up to speed too :-)

GraphML support will probably take more than two weeks as I won't be at 
it full time. I think we need to discuss layouts and decide which ones 
to implement and in what order. From what I've seen when looking at this 
before, is that some algorithms are just variation on a theme. We could 
look at coding the algorithms in 'families'?

As regards your summary below what are you working on, or more to the 
point what do you want me to do? I need item one implemented for the 
start of this week, but if you've already done some work on it, I don't 
want to duplicate the effort.

BTW: I don't know if you realised but most of the mails you've sent were 
  directly to me. If your intention was to send to the list make you 
send it to net...@li....

Regards,

Martin

Illumineo wrote:
>  
> 
> Morning Martin,
> 
> 
> Docking again. My intention is to drop the Magic lib altogether. The reason
> it's still present in NAF is because it is not re-written yet in function of
> the DockPanel. Not too much work but hasn't been a priority since preview 4
> of NAF for me.
> 
> CVS module aliasing. I'd say: keep it as simple as possible. The graph lib
> is an integral part of NAF and I did quite a bit of efforts to have it
> bi-valent, i.e. a control for everyone and a control inside the framework.
> How much people are interested in using the graph control inside NAF is
> another issue...at least I try to make it work.
> 
> WIP (work in progress) and CVS. You are right I suppose about having things
> compilable if in CVS. Regarding this I have always seen 'releases' as true
> intermediate units and not CVS modules. Whenever I download open source
> stuff I always take releases and never use the CVS but I suppose I've the
> wrong perception. In my professional life people dump code in VSS in any
> state: buggy, finished, half-started...It's an agreement and I tend to think
> that the CVS community and the VSS community are different kind of people.
> This says nothing about values of course, merely an observation. 
> If I can be totally honest with you, I'd rather prefer to leave everything
> CVS to you. 
> 
> New features. I would not mind to integrate your code (or whoever submits to
> Netron) and deliver it as part of V2. You are probalby aware of the
> difference between V1b and V2; undo/redo, intermediate connectors in
> connections, ...I took one step back to go two steps forward. I believe the
> code is now much cleaner than before and I rewrote quite a bit. The custom
> connection where somewhere around in v1b but not well implemented. If you
> have code for GraphML and/or layout, let me know. Try to estimate how much
> time it'll cost, more than two weeks integration is too much. I hope Helge
> will pop up with some VB.Net code but I'm not sure. Recently a funny guy
> said he got Netron working in ASP.Net (see the blog), maybe an example of
> this would be nice too.
> 
> 
> Let me summarize:
> 
> 1. User defined connection, we go for it. 
> 
> 2. User definable context menu on an Entity/Shape, we go for it. 
> 
> 3. Make Entity/Shape mouse aware, we go for it.
> 
> 4. Clean CVS. You take care. A compilable cvs module should now be
> available, the next one will contain the new items above.
> 
> 5. Time estimation of GraphML and layout. Not a must for V2.
> 
> 
> 
> What and how do you use the graph lib by the way (curious about the bias)?
> In a professional context?
> 
> I am writing a technical paper for the moment about the lib, not much coding
> these days. Well, that's not true; I am coding but not related to Netron.
> 
> 
> 
> Have a nice weekend!
> 
> 

Re: [Netron-graphlib] Some clarifications

[Martin C. <mar...@ei...>]

Hi Francois,

I have successfully built all projects from the ZIP file except 
NAFBaseServices. I think this is due to it needing some stuff that is 
only available when NAFEnabled is defined. I tried compiling with 
NAFEnabled defined but that gives errors also. Removing NAFBaseServices 
allows everything to build. What should I do about this?

Regards,

Martin

[Netron-graphlib] Some clarifications

[Illumineo <Net...@ho...>]

A lot of questions, let me explain. Please don't take my comments as an
offense, I try to state my point of view and how things have been until now,
i.e. until I got your kind help.


About docking. The Magic library has been the default choice for a long time
'cause nothing else was available on the open source market. They change
their licensing and like many I moved to Luo's library. This library moved,
indeed, from a Magic substrate (DockManager) to a DockPanel based approach.
The situation is now; everything NAF uses still the Magic library,
everything graph (i.e. all graph applications) use Luo's lib. This
discrepancy is due to the fact that I cannot work on everything
simultaneously and is not intentional. In the full NAF solution you can see
the project 'MagicLibrary' and the project 'WinFormsUI'. If you want to
compile all projects you need both, if you only want the graph projects the
latter should be sufficient.


About 'work in progress'. Yes, there is a lot of work in progress. I did try
to round up the graph lib stuff though. Besides GraphML serialization and
the rectangular layout algorithm (which was pointed out by Martin, I didn't
believe anyone would notice...), I believe the graph lib is pretty much
self-contained and 'OK' as far as features are concerned. The Netron
framework on the other hand is another story but I don't think this an issue
in the context of this discussion.

About v2 features. I am open to additions if it does not mean a complete
re-writing of the library. I have been rewriting a lot and it is time to let
it go now, for me at least. Many people have asked this-or-that in the past
year and it is totally impossible to have everything in one package. My
vision (mission?) is the following: the Netron library is *not* a Visio
remake, it's not a UML oriented library, it's not a simulation tool...but if
you wish you can take it and model it to your own needs. The things I'm
after can be seen in the Biotron and Automtron applications.
Now, about the user-defined connections and the VB.Net support. I tend to
say 'yes' for it but you need to tell me how long and how deep the mods are.
I would be very glad to add user-connection and at least one sample VB.Net
application, maybe even an ASP.Net sample. 

About CVS. This is an everlasting pain for me. In the past two years, or so,
I've used CVS only because others were asking for it (where's the latest
code?). I'd be very happy if Martin would continue to take care of CVS. On
the other hand I do realize that co-operating on source code doesn't work
without a versioning system. It's difficult to work together even with CVS.

Concerning the last CVS-commit: I did not add any new stuff, I only
committed the sources already present in CVS. I am sorry if this scrambled
the repository, Martin. I'll try to pick up WinCVS again (how many nights
have I had with CVS already...?). Your suggestion of writing something about
Net development and CVS is great I think, please write it down and
illuminate others!


About the mailing list. Thank you for this suggestion, I hope this makes
communicating more efficient! Let me know if more stuff needs improvements.


That's it for now. Hope this clarifies a little. Martin and Helge, can you
give me an idea 'bout the delay/work I mentioned above?

Thanks a lot for you help again, I am grateful to you.

Francois.