[squeak-dev] [Ann] Cuis: A new Squeak distribution

>
> On 28.03.2009, at 13:31, Juan Vuletich wrote:
>
>> Damien Cassou wrote:
>>> On Fri, Mar 27, 2009 at 9:16 PM, Juan Vuletich <[hidden email]>
>>> wrote:
>>>
>>>> I'm pleased to announce that Cuis is available. Cuis is a Squeak
>>>> distribution with emphasis on simplicity. You can download it from
>>>> http://www.jvuletich.org/Cuis/Cuis1.0-0158.zip . The web is not
>>>> done yet.
>>>>
>>>
>>> The content looks very ugly when using a composite window manager (I'm
>>> on Ubuntu with Compiz). When switching off compiz, things looks ok.
>>>
>>>
>> If I remeber correctly, these kind of problems also happened with
>> other Squeak distributions and was due to a kind of bug in BitBlt,
>> right?
>
> Not a BitBlt bug, but how BitBlt is used. If all drawing operations
> would set the Display's alpha component to 1 then it would be fully
> opaque.
>

> Bert Freudenberg wrote:
>>
>> On 28.03.2009, at 13:31, Juan Vuletich wrote:
>>
>>> Damien Cassou wrote:
>>>> The content looks very ugly when using a composite window manager  
>>>> (I'm
>>>> on Ubuntu with Compiz). When switching off compiz, things looks ok.
>>>>
>>> If I remeber correctly, these kind of problems also happened with  
>>> other Squeak distributions and was due to a kind of bug in BitBlt,  
>>> right?
>>
>> Not a BitBlt bug, but how BitBlt is used. If all drawing operations  
>> would set the Display's alpha component to 1 then it would be fully  
>> opaque.
>>
> It is a BitBlt bug. Evaluate and print this:
>
> f1 _ Form extent: 8@8 depth: 32.
> f1 fillColor: (Color r: 1.0 g: 1.0 b: 0.7 alpha: 0.6).
> f2 _ Form extent: 8@8 depth: 32.
> f2 fillColor: (Color r: 0.5 g: 1.0 b: 0.7 alpha: 0.7).
> bb _(BitBlt toForm: f1) sourceForm: f2; combinationRule: 28; copyBits.
> f1 display.
> (f1 colorAt: 2@2) alpha
>
> Rule 28 is rgbMin. The resulting color should be (Color r: 0.5 g:  
> 1.0 b: 0.7 alpha: 0.6). But alpha is set to zero. This is a bug, and  
> it is what causes wrong alpha values in Display in Cuis.

> Comments below.
>
> On 30.03.2009, at 00:44, Juan Vuletich wrote:
>
>> It is a BitBlt bug. Evaluate and print this:
>>
>> f1 _ Form extent: 8@8 depth: 32.
>> f1 fillColor: (Color r: 1.0 g: 1.0 b: 0.7 alpha: 0.6).
>> f2 _ Form extent: 8@8 depth: 32.
>> f2 fillColor: (Color r: 0.5 g: 1.0 b: 0.7 alpha: 0.7).
>> bb _(BitBlt toForm: f1) sourceForm: f2; combinationRule: 28; copyBits.
>> f1 display.
>> (f1 colorAt: 2@2) alpha
>>
>> Rule 28 is rgbMin. The resulting color should be (Color r: 0.5 g: 1.0
>> b: 0.7 alpha: 0.6). But alpha is set to zero. This is a bug, and it
>> is what causes wrong alpha values in Display in Cuis.
>
> Not a bug per se, it's called rgbMin and not rgbaMin, right? Besides,
> even it would take the min of the alpha, then the alpha in the frame
> buffer would still not be 1 to indicate opaqueness.
>

> On Fri, Mar 27, 2009 at 5:16 PM, Juan Vuletich <[hidden email]
> <mailto:[hidden email]>> wrote:
>
>     Hi Folks,
>
>     I'm pleased to announce that Cuis is available. Cuis is a Squeak
>     distribution with emphasis on simplicity. You can download it from
>     http://www.jvuletich.org/Cuis/Cuis1.0-0158.zip . The web is not
>     done yet.
>
>     Some of the main ideas and objectives for Cuis are:
>
>     - Aimed at Smalltalk developers and people learning about Smalltalk
>     - Close to Smalltalk-80 and Dan Ingalls' ideas
>     - Compatible with the latest Squeak VMs
>     - MIT license (based on the work by Yoshiki Ohshima, from VPRI)
>     - Evolution by removing unnecessary complexity, not adding it
>     - Reduced Morphic
>     - No MVC or Etoys
>     - No M17N, Traits, Monticello, Omnibrowser, TTFonts, FreeType
>     - Includes support for building VM plugins, but not for building
>     the VM
>     - Includes a set of high quality, antialiased StrikeFonts
>     - Backwards compatibility not important
>     - Include code from the Squeak project (various versions, Mantis,
>     squak-dev, etc)
>
>     Cuis is under MIT license. It is the result of several years of
>     cleanup, and it is currently being used in one commercial project.
>     It is small and fast, and good for dev and learning. Besides PCs,
>     it also runs great on older machines, PDAs and embedded devices.
>
>     I hope you like it.
>
>     Cheers,
>     Juan Vuletich
>
>
> ------------------------------------------------------------------------
>
>
>  
> ------------------------------------------------------------------------
>
>
> No virus found in this incoming message.
> Checked by AVG - www.avg.com
> Version: 8.5.285 / Virus Database: 270.11.32/2030 - Release Date: 03/30/09 08:40:00
>
>  

> On Mon, Mar 30, 2009 at 5:55 PM, Hernan Wilkinson
> <[hidden email] <mailto:[hidden email]>> wrote:
>
>     Hi Juan,
>      great work! a friend just told me and I could not wait to try
>     it.... it flies!!!
>
>     On Fri, Mar 27, 2009 at 5:16 PM, Juan Vuletich <[hidden email]
>     <mailto:[hidden email]>> wrote:
>
>         Hi Folks,
>
>         I'm pleased to announce that Cuis is available. Cuis is a
>         Squeak distribution with emphasis on simplicity. You can
>         download it from
>         http://www.jvuletich.org/Cuis/Cuis1.0-0158.zip . The web is
>         not done yet.
>
>         Some of the main ideas and objectives for Cuis are:
>
>         - Aimed at Smalltalk developers and people learning about
>         Smalltalk
>         - Close to Smalltalk-80 and Dan Ingalls' ideas
>         - Compatible with the latest Squeak VMs
>         - MIT license (based on the work by Yoshiki Ohshima, from VPRI)
>         - Evolution by removing unnecessary complexity, not adding it
>         - Reduced Morphic
>         - No MVC or Etoys
>         - No M17N, Traits, Monticello, Omnibrowser, TTFonts, FreeType
>         - Includes support for building VM plugins, but not for
>         building the VM
>         - Includes a set of high quality, antialiased StrikeFonts
>         - Backwards compatibility not important
>         - Include code from the Squeak project (various versions,
>         Mantis, squak-dev, etc)
>
>         Cuis is under MIT license. It is the result of several years
>         of cleanup, and it is currently being used in one commercial
>         project. It is small and fast, and good for dev and learning.
>         Besides PCs, it also runs great on older machines, PDAs and
>         embedded devices.
>
>         I hope you like it.
>
>         Cheers,
>         Juan Vuletich
>
>
>

>
>
> Cheers,
> Juan Vuletich
>
>> On Mon, Mar 30, 2009 at 5:55 PM, Hernan Wilkinson <[hidden email]
>>  <mailto:[hidden email]>> wrote:
>>
>>   Hi Juan,
>>    great work! a friend just told me and I could not wait to try
>>   it.... it flies!!!
>>
>>   On Fri, Mar 27, 2009 at 5:16 PM, Juan Vuletich <[hidden email]
>>   <mailto:[hidden email]>> wrote:
>>
>>       Hi Folks,
>>
>>       I'm pleased to announce that Cuis is available. Cuis is a
>>       Squeak distribution with emphasis on simplicity. You can
>>       download it from
>>       http://www.jvuletich.org/Cuis/Cuis1.0-0158.zip . The web is
>>       not done yet.
>>
>>       Some of the main ideas and objectives for Cuis are:
>>
>>       - Aimed at Smalltalk developers and people learning about
>>       Smalltalk
>>       - Close to Smalltalk-80 and Dan Ingalls' ideas
>>       - Compatible with the latest Squeak VMs
>>       - MIT license (based on the work by Yoshiki Ohshima, from VPRI)
>>       - Evolution by removing unnecessary complexity, not adding it
>>       - Reduced Morphic
>>       - No MVC or Etoys
>>       - No M17N, Traits, Monticello, Omnibrowser, TTFonts, FreeType
>>       - Includes support for building VM plugins, but not for
>>       building the VM
>>       - Includes a set of high quality, antialiased StrikeFonts
>>       - Backwards compatibility not important
>>       - Include code from the Squeak project (various versions,
>>       Mantis, squak-dev, etc)
>>
>>       Cuis is under MIT license. It is the result of several years
>>       of cleanup, and it is currently being used in one commercial
>>       project. It is small and fast, and good for dev and learning.
>>       Besides PCs, it also runs great on older machines, PDAs and
>>       embedded devices.
>>
>>       I hope you like it.
>>
>>       Cheers,
>>       Juan Vuletich
>>
>>
>>
>
>

>
> On Mar 31, 2009, at 15:12 , Juan Vuletich wrote:
>
>> Hernan Wilkinson wrote:
>>> besides cleaning up Morph a little bit, did you do something else to
>>> make the UI run faster?
>>>
>> Not sure. I removed and cleaned stuff everywhere, so it's hard to say
>> what had impact on UI performance and what didn't.
>
> Hi Juan,
>
> Did you create changesets or MC packages so that your cleanups could
> be reused (e.g., in Pharo or Squeak)?
>
> Adrian

> Bert Freudenberg wrote:
>> On 30.03.2009, at 00:44, Juan Vuletich wrote:
>>
>>> It is a BitBlt bug. Evaluate and print this:
>>>
>>> f1 _ Form extent: 8@8 depth: 32.
>>> f1 fillColor: (Color r: 1.0 g: 1.0 b: 0.7 alpha: 0.6).
>>> f2 _ Form extent: 8@8 depth: 32.
>>> f2 fillColor: (Color r: 0.5 g: 1.0 b: 0.7 alpha: 0.7).
>>> bb _(BitBlt toForm: f1) sourceForm: f2; combinationRule: 28;  
>>> copyBits.
>>> f1 display.
>>> (f1 colorAt: 2@2) alpha
>>>
>>> Rule 28 is rgbMin. The resulting color should be (Color r: 0.5 g:  
>>> 1.0 b: 0.7 alpha: 0.6). But alpha is set to zero. This is a bug,  
>>> and it is what causes wrong alpha values in Display in Cuis.
>>
>> Not a bug per se, it's called rgbMin and not rgbaMin, right?  
>> Besides, even it would take the min of the alpha, then the alpha in  
>> the frame buffer would still not be 1 to indicate opaqueness.
>>
> I believe it is a bug. Reasonable answers would be alpha = 1 (always  
> generate opaque colors), 0.6 (previous destination alpha), 0.7 (max  
> alpha between source and destination) or 0.65 (mean alpha between  
> source and destination). Any of these results would be consistent  
> with the result for other bit depths. Setting alpha to zero means  
> that if the destination depth is 32, rule 28 will always fill  
> destination with transparent, ignoring everything else. I think this  
> can never be the desired result.
>
> BTW, the same problem happens with the following rules:
>   20    rgbAdd: sourceWord with: destinationWord.  Sum of color  
> components
>   21    rgbSub: sourceWord with: destinationWord.  Difference of  
> color components
>   27    rgbMax: sourceWord with: destinationWord.  Max of each color  
> component.
>   28    rgbMin: sourceWord with: destinationWord.  Min of each color  
> component.
>   29    rgbMin: sourceWord bitInvert32 with: destinationWord.  Min  
> with (max-source)
>   37    rgbMul:with:
>
> All these rules are currently useless if destination is 32 bit deep.
>
>> I can't remember what the rgbMin rule is used for. Changing it  
>> might break something, or it might now. Does anyone know?
>>
> It should break nothing. Nobody can be actually using that  
> transparent in a meaningful way.

>
>> ...
>>> I can't remember what the rgbMin rule is used for. Changing it might
>>> break something, or it might now. Does anyone know?
>>>
>> It should break nothing. Nobody can be actually using that
>> transparent in a meaningful way.
>
>
> Not for display purposes. But BitBlt has been used for numeric stuff,
> too, where it would make a difference.
>
> OTOH I agree that making the results in 32 bits match the ones in
> lower bit depths is sensible. Care to send a changeset to vm-dev?
>
> - Bert -
>
>