On Monday, September 29, 2003, at 05:46 PM, Mike Roberts wrote:

> I was playing around in the image testing things and I noticed 
> something odd with URL links.  I'm not sure if this has been noticed 
> before.  It's not show-stopping, just a little niggly.
>
> The 'Welcome to...' window provides some examples.
>
> The first thing isn't odd, its just that the 'VM' link is a TextDoIt 
> and directly loads Scamper.  This is great.  However, the 'Basic' link 
> is a TextURL and this requires a confirmation click.  Apart from being 
> inconsistent, it seems a shame to have to confirm the link when 
> Scamper is available.

Good point, I created the 'Basic' and 'Full' links more recently than 
the 'VM' link, and I guess they are different.  TextURL would seem like 
the right type of link for this, actually.  I just tried testing the 
'VM' link in the 3.6gamma1-basic image (which doesn't include Scamper), 
and it gets a MessageNotUnderstood.  Probably the TextURL behavior 
should be improved so that it doesn't bother prompting you to open a 
browser if one is already present.

> I have had a quick look at TextURL and in actOnClickFor: it seems to 
> be doing quite a lot.  I'm not sure what people think about this.  
> That aside, here is the odd behaviour.
>
> Sometimes, when you move the mouse the dialog is automatically 
> confirmed.  Sometimes it is automatically canceled.  I'm not sure 
> where this is happening but I am not clicking on either of the options 
> so I'm not sure if an event is being generated somewhere.

Hmm, I think I've also seen this odd behavior once or twice with this 
dialog, but I'm not able to get it to happen reliably right now.  (I 
might have seen it only on Windows... I wouldn't be surprised if it was 
machine/performance dependent...)

> Also, if you really don't want to view the link then you have to 
> confirm that you have cancelled as the method offers you the source.  
> The comment in the code indicating the intent is fair enough if it 
> really couldn't find a web browser, but that isn't the case in this 
> instance.  The logic probably just needs to be tidied up here.

Yes.  Patches welcome. ;-)

> I'm not sure how best to debug these links.  Is there a quick way to 
> get up the contents of the special action?  I resorted to inspecting 
> through the run array and finding the correct Text objects for each 
> one.

To see the contents of the special action, you need to highlight the 
entire link first, then bring up the alt-6 menu and select "Edit hidden 
info".  Select "be a weburl link" to change it back to a TextURL.

(Yes, the alt-6 menu is pretty obscure... it should be made accessible 
via the mouse only if needed.)

> In trying the actions out in the inspector, I just did a self 
> actOnClickFor:self and probably obviously self is not the same as in 
> the Welcome window as self is the (inspector) TextURL.  The mouse 
> funnies don't happen here.  By inserting a halt in the method, I can 
> see that the object passed to actForClickOn: in 'Welcome to...' is a 
> Workspace. hmm..  There isn't anything special in this case so I 
> wonder what it is about the Welcome window that maybe interacts with 
> the pop up menus?  This approach to debugging this seems a bit bogus 
> anyway because it has a feel of a timing issue or something to do with 
> the mouse events and sticking in a halt has the feeling of stopping 
> Morphic in its tracks.  I don't know enough about morphic to know 
> really what happens here - have you just stopped one specific step?

I don't have any good answers here except to say that yes, this sort of 
thing is difficult to debug. :-)

> I've just had another play around and the best way to provoke it is to 
> click on 'Basic' and don't move the mouse.  The hand is just on the 
> left edge of the pop-up.  Immediately move the hand to the left and 
> you get a 'cancel'.  Or.  Immediately move the mouse right into 'Yes' 
> and you get a yes.  Or.  Immediately move the house up into the 
> heading bar of the menu and then down into 'yes' and it behaves 
> normally.  Weird :-)
>
> Anyway I'm rambling.  Someone may spot what this is.
>
> Apart from that, I had a great time playing around in Worlds of 
> Squeak.  I didn't manage to break anything, which is unusual :-)  Oh, 
> there was one thing.  There is a bug of sorts in the Freecell game.  
> When you click on a card, the morphic dragging method does some 
> thinking about whether you want to pick up a stack etc.  If you move 
> the mouse quickly, you can get the mouse away from the cards before 
> the cards have started to move.  The drag point is then in the wrong 
> place.  If you wait a little while before moving, the cards move 
> correctly under the hand.  I'm not sure what is happening here.  Any 
> ideas?

Not sure exactly, except that it's probably event-related.  Also, I'd 
guess this bug has probably been there quite awhile, probably since 3.0 
or maybe it's always been there.  So, it won't likely get fixed for 
3.6. :-)  I think I have seen this behavior myself, too... I'm guessing 
it only shows up on somewhat slower machines?

Anyway, thanks for all of the testing.

- Doug