This is because drawview (incorrectly, IMO) is thinking that it knows the 
internal representation of colors.

I'll have a change for you to review in a minute.

On 2010-11-16, at 13:19, Max Carlson wrote:

> See http://jira.openlaszlo.org/jira/browse/LPP-8778 for a testcase...
> 
> On 11/16/10 10:15 AM, P T Withington wrote:
>> Can you provide a small test case?  I see no pink box...
>> 
>> On 2010-11-16, at 12:40, Max Carlson wrote:
>> 
>>> I just found a regression in swf10 :(.  See the pink box in 
>>> http://localhost:8080/trunk-clean/test/drawing/drawing.lzx?lzr=dhtml&lzt=html
>>>  vs 
>>> http://localhost:8080/trunk-clean/test/drawing/drawing.lzx?lzr=swf10&lzt=html.
>>>   This has to do with the way the alpha color is processed in drawview...
>>> 
>>> 
>>> On 11/15/10 2:53 PM, P T Withington wrote:
>>>> Change ptw-20101115-YPt by [email protected] on 2010-11-15 17:28:26 EST
>>>>     in /Users/ptw/OpenLaszlo/trunk-devo
>>>>     for http://svn.openlaszlo.org/openlaszlo/trunk
>>>> 
>>>> Summary: Change internal color representation to encode alpha accurately
>>>> 
>>>> Bugs Fixed:
>>>> LPP-9154 Investigate using 32bit ints instead of floats in LzColorUtils
>>>> LPP-9528 LzColorUtils does not handle rbga(n,n,n,0) correctly
>>>> 
>>>> Technical Reviewer: [email protected], [email protected] (pending)
>>>> QA Reviewer: [email protected] (pending)
>>>> 
>>>> Release Notes:
>>>> 
>>>>     Colors are now represented internally as 32-bit integers with any
>>>>     alpha value being stored as its complement in the high 8 bits.
>>>>     Thus, a color with an alpha of 1 will be represented by the 24-bit
>>>>     color as before, for compatibilty with code that does not
>>>>     understand alpha.
>>>> 
>>>>     Code that intends to respect alpha needs to be aware of the alpha
>>>>     component of colors and be sure to preserve it when manipulating
>>>>     colors.  This is best done by unpacking the internal representaion
>>>>     into components, manipulating the components, and then repacking
>>>>     them.  All standard color models are supported for unpacking and
>>>>     repacking -- choose the one best suited to your manipulations.
>>>> 
>>>> Overview:
>>>> 
>>>>     Basically implemented a whole new substrate in LzColorUtils that
>>>>     knows how to pack/unpack internal representations from/to any of
>>>>     rgba, hsla, hsva, 8-digit # colors, etc.  Wrote a much more
>>>>     complete CSS colorspec parser and unparser.  Shoved these
>>>>     underneath the existing API's in as compatible a manner as
>>>>     possible.
>>>> 
>>>> Details:
>>>>     lzunit-lzutils: Updated tests that depended on internal
>>>>     representation, which has changed.
>>>> 
>>>>     LzUtils: New color representation, pack/unpack methods, css
>>>>     colorspec parser and unparser.  Use those as the substrate for the
>>>>     existing color APIs.
>>>> 
>>>>     style: Bug discovered by new code -- the color type auto-quotes
>>>>     its value property.  If you want the value to be an expression,
>>>>     you must use ${}.
>>>> 
>>>> Tests:
>>>>     {smokecheck, lzuint-lzutils} x {swf8, swf10, dhtml}
>>>> 
>>>> Files:
>>>> M       test/lfc/lzunit-lzutils.lzx
>>>> M       WEB-INF/lps/lfc/services/LzUtils.lzs
>>>> M       lps/components/base/style.lzx
>>>> 
>>>> Changeset: 
>>>> http://svn.openlaszlo.org/openlaszlo/patches/ptw-20101115-YPt.tar
>> 


Reply via email to