Mouse over connection when scrolled away from top left on page.

[GoogleCodeExporter]

I'm using v 1.3.10 of jsPlumb and the recommended other libraries from your 
demo pages - I'm basically encoding the connection data required into JSON in 
.net and then creating connections based on that data.

The diagrams are dynamic and some scroll off the bottom of the screen - so it's 
the screen that scrolls rather than the div inside the window - at any point if 
I scroll the page down or to the right it draws any connections that the mouse 
makes contact with in the wrong place on mouseexit.  I've tried to illustrate 
with diagrams - it draws them in the wrong place consistently offset with how 
far from top left you have scrolled.

I hope you can help because I'm in the middle of a solution that uses JSPlumb 
and this would be a blocker to be able to use it.

I did see other posts about what seemed like a similar problem but couldn't 
find anything that fixed it for me - repaintEverything makes things 
considerably worse and other forum posts seem to indicate it is fixed.

Thanks
Steve

Original issue reported on code.google.com by steswork...@gmail.com on 25 Jun 2012 at 3:48

Attachments:

• [on screen.png](https://storage.googleapis.com/google-code-attachments/jsplumb/issue-259/comment-0/on screen.png)
• [off screen.png](https://storage.googleapis.com/google-code-attachments/jsplumb/issue-259/comment-0/off screen.png)

[GoogleCodeExporter]

i definitely would like to make sure that issues like this are fixed in 
jsPlumb, but it's hard for me to diagnose the issue without a test page.  could 
you maybe create something on jsFiddle that recreates the problem?

Original comment by simon.po...@gmail.com on 25 Jun 2012 at 10:28

[GoogleCodeExporter]

During my creation of a test page locally to know how it would look I think I 
found the problem - while I was prototyping I had draggable set from JQuery-UI 
and not jsPlumb - my test page was using just the jsPlumb code and I set 
draggable in there and couldn't reproduce the problem so I looked for what was 
different which involved changing how I did draggable - and that seemed to fix 
it.  So it gets confused if you use draggable from ui rather than jsPlumb - my 
mistake - sorry for and thanks for your time!

Original comment by steswork...@gmail.com on 26 Jun 2012 at 10:54

[GoogleCodeExporter]

This all seemed to have been fixed but I am having the same problems again 
today - seeing as it still seems to be open - the code I am using is here:

http://jsfiddle.net/UxcjM/ 

This appears to work in jsFiddle but does not work at all in IE8 whether or not 
the excanvas library is there or not - I've not added excanvas into the fiddle 
because I can't see it in the source of your demos.

I'm not saying it's definitely a defect but would appreciate some feedback if 
it's my mistake.

Thanks

Original comment by steswork...@gmail.com on 3 Jul 2012 at 1:49

[GoogleCodeExporter]

Oh and one other thing I noticed - one of the endpoints (think it's only one) 
gets drawn in the top corner of the screen - that's there in JSFiddle as well - 
that one's just cosmetic for me at the moment compared to the larger issue.

Original comment by steswork...@gmail.com on 3 Jul 2012 at 2:18

[GoogleCodeExporter]

Sorry URL should be http://jsfiddle.net/UxcjM/1/ - sorry bit new to this.

Original comment by steswork...@gmail.com on 3 Jul 2012 at 2:32

[GoogleCodeExporter]

i'll take a look at this in a couple of days - i'm a little busy at the moment. 
 if it's a bug i will fix it of course.  i'm sure it won't take long to fix.

excanvas is not a jsplumb dependency.

Original comment by simon.po...@gmail.com on 5 Jul 2012 at 6:54

[GoogleCodeExporter]

That's all I can ask - thanks in advance for taking a look!

Original comment by steswork...@gmail.com on 5 Jul 2012 at 10:11

[GoogleCodeExporter]

hi

i finally got around to taking a look at this - in the jsfiddle i dont see any 
problem.  which i think you mentioned.  wel, in fact i do see an endpoint at 
0,0, which is not good, but i don't see any paint issues like in your 
screenshot.  so i will investigate the endpoint, but without a recreation of 
the problem you are seeing i cannot help you with the other bit.

Original comment by simon.po...@gmail.com on 25 Jul 2012 at 12:08

[GoogleCodeExporter]

ok i've boiled down the fiddle you posted to this:

http://jsfiddle.net/qpmwv/25/

i suspect that the misplaced endpoint is caused by whatever is causing the 
misplaced connectors for you. i noticed there are nested tables; perhaps this 
is somehow the cause.  this fiddle works fine in chrome.

i will try to fix this but i don't think it will make it into my next release, 
which is happening in the next couple of days.

Original comment by simon.po...@gmail.com on 25 Jul 2012 at 12:31

[GoogleCodeExporter]

i've looked further at this...you might not like what i have concluded.  i 
don't think i'm going to try to fix it.  it's really a combination of how much 
spare time i have along with the convoluted markup and the fact that its only 
older IE.  

Original comment by simon.po...@gmail.com on 25 Jul 2012 at 10:04

• Changed state: WontFix

[GoogleCodeExporter]

OK well I understand - can't say I'm not disappointed.  Basically the use to 
which I'm putting JSPlumb is to describe network topology - the cables can 
connect to 4 different types of endpoints within a hierarchy and it's important 
to the user to be able to distinguish them.  With you saying that the markup is 
convoluted - would there be any suggestions you would have to achieve the same 
thing in a different way?  Using different elements or any suggestions like 
that?  I was looking at maybe connecting them at an absolute position on the 
main table in which case I wouldn't need to connect to the nested table - or if 
it was a div instead - or any suggestions at all would be helpful

Thanks anyway.

Original comment by steswork...@gmail.com on 2 Aug 2012 at 7:50

[GoogleCodeExporter]

I know what you mean - it seems like a shame to leave something broken.  I'm 
not a fan of that either.

In this case it's a combination of the fact that we're talking about old IE and 
nested tables. Tables are something that IE handles in a notoriously fickle 
way, and the presence of the Form element enclosing everything also puts me 
somewhat on IE alert.  This, along with the fact that it works in modern 
browsers, and also how busy I find myself to be these days, all combine to make 
this something I'm reluctant to put any time into.

But if you are open to suggestions on the markup - have you thought about 
replacing all those tables with unordered lists?  Without seeing your app its 
hard for me to know with certainty, but you could definitely achieve what's in 
that fiddle just using ULs.

Original comment by simon.po...@gmail.com on 2 Aug 2012 at 11:41

[GoogleCodeExporter]

Hmm - well I'm more of a server side developer as a rule so I don't dabble much 
in the markup and usually leave it to a Web Designer - I'm not sure I'll have 
time to change it for the first iteration but I could do a bit of research and 
see whether that would work and put it in version 2 if so - I was definitely 
open to the suggestion that I wasn't necessarily doing it the best way.  So are 
you suggesting there that I could connect up the LI elements then?

Original comment by steswork...@gmail.com on 6 Aug 2012 at 1:07

[GoogleCodeExporter]

you can connect up any elements you like.  well of course the existence of this 
issue kind of puts the lie to that statement, but in theory you can connect any 
elements you like, and in practice this only falls down when you get into shaky 
territory on a shaky browser - which is what I consider two nested tables 
inside a form to be in IE8.  

ULs, to me, look like a more natural choice for the markup here. But then I am 
only working off the fiddle you posted.  I'm not completely anti-tables as some 
people are.  I like them, for, well, tabular data.  Ha.  

Original comment by simon.po...@gmail.com on 8 Aug 2012 at 7:00