Author Topic: Label attachment links (Read 5720 times)

Paolo F Cantoni · « **on:** July 31, 2007, 09:06:28 pm »

The new label attachment links (that tell you which edge your label is linked to) are cool!

However, they would be even cooler if they could be optionally rendered onto the diagram.

That way, you could layout things a bit more clearly but still know which label went with which edge.

Thoughts? Votes?
Paolo

Paolo F Cantoni · « **Reply #1 on:** July 31, 2007, 09:11:24 pm »

They would be even cooler if the rendered attachment links could be positioned at each end...

Paolo

thomaskilian · « **Reply #2 on:** July 31, 2007, 10:23:45 pm »

Maybe, but could also lead to cluttering the image with lots of dashlines. Then you likely have the need to turn it on/off on a per-line basis which seems to be tedious. I recently worked with component diagrams with lots of interfaces. I recognized that the label was set of the interface so far that it appeared to be attached to the neighbour. No idea what could be a solution. :-/

Paolo F Cantoni · « **Reply #3 on:** July 31, 2007, 10:45:56 pm »

Yes, Thomas, that's exactly the point! You balance the tedium of enabling/disabling each attachment link against the amount of clutter. At present, You have no choice...

Paolo

«Midnight» · « **Reply #4 on:** August 01, 2007, 01:23:07 am »

Yes indeed,

All things considered I'd go with Paolo on this one. Not having any say is getting to be a drag.

Perhaps there could be a global option to disable these, so situations where there are a lot of busy diagrams don't get too onerous.

David

Paolo F Cantoni · « **Reply #5 on:** August 01, 2007, 02:12:01 am »

Quote

Perhaps there could be a global option to disable these, so situations where there are a lot of busy diagrams don't get too onerous.

David,

To clarify... I had envisaged having to display/hide each attachment individually (the default would be "hide"). Most of the time, displaying these attachments would be unnecessary. They're really only necessary for cluttered diagrams where you can "call-out" the label from the edge. I can't see any point in globally displaying ALL the attachments (or have I missed something).

Paolo

«Midnight» · « **Reply #6 on:** August 01, 2007, 02:24:30 am »

No Paolo, you have not.

I was merely 'clarifying' that there has to be some way to control the default. That would allow users to 'customize' the overall behavior of EA as required.

The example I gave muddies the water a bit. It would handle Thomas' situation where cluttered diagrams were an issue. However, some users might want to turn off the function in its entirety. Others (myself included) would want it just the way you describe.

David

thomaskilian · « **Reply #7 on:** August 04, 2007, 02:06:59 pm »

Just an insertion, Paolo: On the one hand you claim CCC and on he other hand you poke into EA to widen up making it more NCNCNC. I see the demand for more functionality, but, but, but...

Paolo F Cantoni · « **Reply #8 on:** August 04, 2007, 07:18:59 pm »

Quote

Just an insertion, Paolo: On the one hand you claim CCC and on he other hand you poke into EA to widen up making it more NCNCNC. I see the demand for more functionality, but, but, but...

Actually, I don't believe so....

I presume, by NC you mean rendering of attachment links, permanently?

The [size=13]UML 2.1.1 Superstructure (formal)[/size] Specification is (unsurprisingly) very quiet about how you manage REAL diagrams with real amounts of information on them; rather than the fairly trite examples on most publications. It doesn't take too long before one is hard-pressed to find suitable space for the various text objects.

I'm just trying to make the diagrams easier to read. Sparx put the functionality there for the user "on the screen" - I'd just like it "on the paper" also.

CCC ^TM

Paolo

thomaskilian · « **Reply #9 on:** August 05, 2007, 05:54:33 am »

Hi Paolo,
not really my point. NCNCNC = (Non-Consistent)**3. I should have been a bit more clear in my abbreviations.

Paolo F Cantoni · « **Reply #10 on:** August 05, 2007, 02:13:45 pm »

Quote

Hi Paolo,
not really my point. NCNCNC = (Non-Consistent)**3. I should have been a bit more clear in my abbreviations.

I got your meaning of the abbreviation, but obviously not what you were referring to... (I thought you were suggesting the label attachment links themselves were NC - which is true, to a point)

Where exactly was NC?

Intrigued,
Paolo

thomaskilian · « **Reply #11 on:** August 06, 2007, 02:34:49 am »

Paolo, let me try to put it in another way. As you know I am in line with you regarding CCC and I think Sparx shall spend a LOT of time to achieve a bit more (maybe CC at least). Now, on the other hand, each time you add functions to EA you definitely won't make the situation better. It's more the opposite way. So (although I like many of your suggestions) requesting increased functionality contradicts indirectly your own CCC paradigm. Even if the new functionality would be CCC within itself, the fundaments of the building would still be NCNCNC - and refurbishment would be paused in favor of new functionality. So my insertion stands: what do you prefer? CCC or new functions

Paolo F Cantoni · « **Reply #12 on:** August 06, 2007, 05:02:16 am »

Thomas,

I take your point, and stand condemned as accused...

However, in my defence, I (we) don't seem to be getting too far in CCC and the new functions keep coming - so at least make them REALLY useful...

I will observe, in addition, that using Agile principles, every addition of new functionality offers the opportunity to refactor existing code so it works better....

Besides if we can get the Sparxians to think more about useful instead of unique interfaces we may make "small steps" toward the "EA" we all seek...

Paolo

thomaskilian · « **Reply #13 on:** August 06, 2007, 09:09:49 am »

Yep. Let's see whether that works. There's no exception from the rule, only the rule of exception

Sparx Systems Forum

News: