Views with non-integer width and height

Hi Max,

I'm tracking down the problem of tabs with missing borders in dhtml FF
1.5 and 2.0. It looks like it is caused by non-integer values for the
width and/or height of the view. I didn't see any other controls with
this problem but it sounds like a generic issue. The clipping
rectangle built by LzSprite.js has non-integer values.

This isn't a problem in swf or IE.

This is a simple example to demonstrate:

<canvas>
<simplelayout spacing="10"/>

<tabs x="10" height="111.5" width="200">
<tabpane>Tab 1<text>Height = 111.5</text></tabpane>
</tabs>

<tabs x="10" height="111" width="200.5">
<tabpane>Tab 2<text>Width = 200.5</text></tabpane>
</tabs>

<tabs x="10" width="101" height="101">
<tabpane>Tab 3<text>Integer values</text></tabpane>
</tabs>

</canvas>

Any thoughts?

Re: Views with non-integer width and height

Hi Max,

I forgot to send the change I'm trying out:

LzSprite.prototype.__updateClip = function() {
...
        this.__LZdiv.style.clip = 'rect(0px ' + Math.floor(this.width)
+ 'px ' + Math.floor(this.height) + 'px 0px)';
...
}

Phil

On 2/1/07, Philip Romanik <promanik <at> laszlosystems.com> wrote:
> Hi Max,
>
> I'm tracking down the problem of tabs with missing borders in dhtml FF
> 1.5 and 2.0. It looks like it is caused by non-integer values for the
> width and/or height of the view. I didn't see any other controls with
> this problem but it sounds like a generic issue. The clipping
> rectangle built by LzSprite.js has non-integer values.
>
> This isn't a problem in swf or IE.
>
> This is a simple example to demonstrate:
>
> <canvas>
> <simplelayout spacing="10"/>
>
> <tabs x="10" height="111.5" width="200">

Re: Views with non-integer width and height

Perhaps we should use Math.round() instead of Math.floor() so we get the 
extra pixel?  Also, it seems like we should be catching this earlier, 
during setWidth/Height...

-Max

Philip Romanik wrote:
> Hi Max,
> 
> I forgot to send the change I'm trying out:
> 
> 
> LzSprite.prototype.__updateClip = function() {
> ...
>        this.__LZdiv.style.clip = 'rect(0px ' + Math.floor(this.width)
> + 'px ' + Math.floor(this.height) + 'px 0px)';
> ...
> }
> 
> 
> Phil
> 
> 
> On 2/1/07, Philip Romanik <promanik <at> laszlosystems.com> wrote:
>> Hi Max,
>>
>> I'm tracking down the problem of tabs with missing borders in dhtml FF
>> 1.5 and 2.0. It looks like it is caused by non-integer values for the
>> width and/or height of the view. I didn't see any other controls with
>> this problem but it sounds like a generic issue. The clipping

for review, add debugger warning on dereference of null pointer in xpath query

Change 20070201-hqm-4 by hqm <at> IBM-112E6A12815 on 2007-02-01 14:42:17 EST
    in /cygdrive/c/users/hqm/openlaszlo/branches/legals

Summary: check for derefencing null value in xpath query

New Features:

Bugs Fixed:

Technical Reviewer: max (pending)
QA Reviewer: (pending)
Doc Reviewer: (pending)

Documentation:

Release Notes:

Details:

I came across this condition which stopped some app in DHTML, I believe it was when
getting Amazon demo running.

I'm not certain if this is the right level of abstraction to be catching this error,
but I don't want people's DHTML apps to grind to a halt over this...

Tests:

Sorry for lack of test case

Files:
C      WEB-INF/lps/lfc/data/LzDatapointer.lzs


--
Henry Minsky
Software Architect
hminsky <at> laszlosystems.com

for review, test for dereferencing null initialnodes in lazy replication manager

Re: for review, add debugger warning on dereference of null pointer in xpath query

The file looks like it has a conflict (C next to it).  Also, you forgot 
the attachment :)

Henry Minsky wrote:
> Change 20070201-hqm-4 by hqm <at> IBM-112E6A12815 on 2007-02-01 14:42:17 EST
>    in /cygdrive/c/users/hqm/openlaszlo/branches/legals
> 
> Summary: check for derefencing null value in xpath query
> 
> New Features:
> 
> Bugs Fixed:
> 
> Technical Reviewer: max (pending)
> QA Reviewer: (pending)
> Doc Reviewer: (pending)
> 
> Documentation:
> 
> Release Notes:
> 
> Details:
> 
> I came across this condition which stopped some app in DHTML, I believe it
> was when
> getting Amazon demo running.
> 
> I'm not certain if this is the right level of abstraction to be catching
> this error,
> but I don't want people's DHTML apps to grind to a halt over this...
> 
> Tests:
> 
> Sorry for lack of test case 
> 
> Files:
> C      WEB-INF/lps/lfc/data/LzDatapointer.lzs
> 
> 

--

-- 
Regards,
Max Carlson
OpenLaszlo.org

Re: for review, test for dereferencing null initialnodes in lazy replication manager

Do you have a test case?

Henry Minsky wrote:
> Change 20070201-hqm-2 by hqm <at> IBM-112E6A12815 on 2007-02-01 14:54:06 EST
>    in /cygdrive/c/users/hqm/openlaszlo/branches/legals/WEB-INF/lps/lfc/data
> 
> Summary: test for null initialnodes attribute
> 
> New Features:
> 
> Bugs Fixed:
> 
> Stops DHTML error from dereferencing null if initialnodes is null (should
> this never happen?)
> 
> Technical Reviewer: max (pending)
> QA Reviewer: (pending)
> Doc Reviewer: (pending)
> 
> Documentation:
> 
> Release Notes:
> 
> Details:
> 
> 
> 
> 
> Tests:
> 
> Files:
> M      LzLazyReplicationManager.lzs
> 
> Changeset: http://svn.openlaszlo.org/openlaszlo/patches/20070201-hqm-2.tar
> 

--

-- 
Regards,
Max Carlson
OpenLaszlo.org

Re: Views with non-integer width and height

If you round it up the error still remains. Doing a floor() is
consistent with what LzSprite.CSSDimension() does.

Phil

Perhaps we should use Math.round() instead of Math.floor() so we get
the extra pixel?
Also, it seems like we should be catching this earlier, during
setWidth/Height...

-Max

Philip Romanik wrote:

[Hide Quoted Text]
Hi Max,

I forgot to send the change I'm trying out:

LzSprite.prototype.__updateClip = function() {
...
        this.__LZdiv.style.clip = 'rect(0px ' + Math.floor(this.width)
+ 'px ' + Math.floor(this.height) + 'px 0px)';
...
}

Phil

On 2/1/07, Philip Romanik <promanik <at> laszlosystems.com> wrote:

Hi Max,

I'm tracking down the problem of tabs with missing borders in dhtml FF
1.5 and 2.0. It looks like it is caused by non-integer values for the
width and/or height of the view. I didn't see any other controls with
this problem but it sounds like a generic issue. The clipping
rectangle built by LzSprite.js has non-integer values.

This isn't a problem in swf or IE.

This is a simple example to demonstrate:

<canvas>
<simplelayout spacing="10"/>

<tabs x="10" height="111.5" width="200">
<tabpane>Tab 1<text>Height = 111.5</text></tabpane>
</tabs>

<tabs x="10" height="111" width="200.5">
<tabpane>Tab 2<text>Width = 200.5</text></tabpane>
</tabs>

<tabs x="10" width="101" height="101">
<tabpane>Tab 3<text>Integer values</text></tabpane>
</tabs>

</canvas>

Any thoughts?

Thanks!

Phil

--

-- 
Regards,
Max Carlson
OpenLaszlo.org

For Review: Change 20070201-maxcarlson-n Summary: Hand merge changes from amazon/ to amazon-soap/

Change 20070201-maxcarlson-n by maxcarlson <at> max-carlsons- 
computer.local on 2007-02-01 12:51:45 PST
     in /Users/maxcarlson/openlaszlo/legals/demos/amazon-soap

Summary: Hand merge changes from amazon/ to amazon-soap/

New Features: Fixed credit card editing screen

Bugs Fixed:

Technical Reviewer: jcrowley
QA Reviewer: hminsky
Doc Reviewer: (pending)

Documentation:

Release Notes:

Details: Only cosmetic changes.

Tests: http://localhost:8080/legals/demos/amazon-soap/amazon.lzx -  
tested search, details, credit card editing screens

Files:
M      shoppinglist.lzx
M      creditcard.lzx
M      address.lzx
M      amazon.lzx
M      resources.lzx
M      classlib.lzx

Changeset: http://svn.openlaszlo.org/openlaszlo/patches/20070201- 
maxcarlson-n.tar

Re: Views with non-integer width and height

You're right.  Can you send out a changeset and consider it approved?  I 
tested it and it definitely fixes the problem in firefox 2 mac and windows.

Philip Romanik wrote:
> If you round it up the error still remains. Doing a floor() is
> consistent with what LzSprite.CSSDimension() does.
> 
> Phil
> 
> 
> 
> 
> 
> 
> Perhaps we should use Math.round() instead of Math.floor() so we get
> the extra pixel?
> Also, it seems like we should be catching this earlier, during
> setWidth/Height...
> 
> -Max
> 
> Philip Romanik wrote:
> 
> 
> [Hide Quoted Text]
> Hi Max,
> 
> I forgot to send the change I'm trying out:
> 
> 
> LzSprite.prototype.__updateClip = function() {
> ...
>        this.__LZdiv.style.clip = 'rect(0px ' + Math.floor(this.width)
> + 'px ' + Math.floor(this.height) + 'px 0px)';
> ...
> }
> 
> 
> Phil
> 
> 
> On 2/1/07, Philip Romanik <promanik <at> laszlosystems.com> wrote:
> 
> Hi Max,
> 
> I'm tracking down the problem of tabs with missing borders in dhtml FF
> 1.5 and 2.0. It looks like it is caused by non-integer values for the
> width and/or height of the view. I didn't see any other controls with
> this problem but it sounds like a generic issue. The clipping
> rectangle built by LzSprite.js has non-integer values.
> 
> This isn't a problem in swf or IE.
> 
> This is a simple example to demonstrate:
> 
> <canvas>
> <simplelayout spacing="10"/>
> 
> <tabs x="10" height="111.5" width="200">
> <tabpane>Tab 1<text>Height = 111.5</text></tabpane>
> </tabs>
> 
> <tabs x="10" height="111" width="200.5">
> <tabpane>Tab 2<text>Width = 200.5</text></tabpane>
> </tabs>
> 
> <tabs x="10" width="101" height="101">
> <tabpane>Tab 3<text>Integer values</text></tabpane>
> </tabs>
> 
> </canvas>
> 
> Any thoughts?
> 
> Thanks!
> 
> Phil
> 
> 

--

-- 
Regards,
Max Carlson
OpenLaszlo.org