issue 2512 - PNG Tests failure

> Hmmm, weren't these the tests that failed in 1.0 if you ran all the tests in one go? I know they were in the neighbourhood anyway, since I spent some time trying to chase them down, and recognise those method names from then.
>
> Maybe Torsten knows a bit more about them, since he was the one that put me on the trail...
>
> --
> Cheers,
> Peter
>
> 2010/6/5 Fabrizio Perin <[hidden email]>
> http://code.google.com/p/pharo/issues/detail?id=2512
>
> the following tests fail:
>
>
> PNGReadWriterTest>>testPngEncodingColors32
> PNGReadWriterTest>>testPngDecodingColors32
>
> They were both green in Pharo 1.0. Also the current implementation of PNGReadWriterTest class has been file in into the last Squeak trunk image and they are green. So the last modifications on the the PNG package cause those methods fail.
>
> Cheers,
>
> Fabrizio
>
> _______________________________________________
> Pharo-project mailing list
> [hidden email]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
> _______________________________________________
> Pharo-project mailing list
> [hidden email]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project

>  On 05.06.2010 20:10, Oscar Nierstrasz wrote:
>> No, we have other tests that fail only when running all tests --
>> those that test finalization.
>>
>> The PNG tests fail reliably in 1.1.  Unfortunately we were unable to
>> narrow down the cause.
>>
>> There appears to be an off by one error somewhere in encoding (or
>> decoding) the Color black.
>>
>> - on
> Related to:
> http://code.google.com/p/pharo/issues/detail?id=2486
> perhaps?
> Essentially, 32 bit transparent black used to be 0(r) 0(g) (1)b (0)a
> instead of 0(r) 0(g) 0(b) 0(a).
> Which, was wrong. (And caused trouble when using transparent black as
> one of the colors in a bitblt operation)
>
> Cheers,
> Henry

> On 06.06.2010 03:18, Henrik Sperre Johansen wrote:
>> On 05.06.2010 20:10, Oscar Nierstrasz wrote:
>>> No, we have other tests that fail only when running all tests -- those that test finalization.
>>>
>>> The PNG tests fail reliably in 1.1.  Unfortunately we were unable to narrow down the cause.
>>>
>>> There appears to be an off by one error somewhere in encoding (or decoding) the Color black.
>>>
>>> - on
>> Related to:
>> http://code.google.com/p/pharo/issues/detail?id=2486
>> perhaps?
>> Essentially, 32 bit transparent black used to be 0(r) 0(g) (1)b (0)a instead of 0(r) 0(g) 0(b) 0(a).
>> Which, was wrong. (And caused trouble when using transparent black as one of the colors in a bitblt operation)
>>
>> Cheers,
>> Henry
> or  (1)a 0(r) 0(g) (0)b instead of 0(a) 0(r) 0(g) 0(b) / (0)a 0(r) 0(g) (1)b instead of 0(a) 0(r) 0(g) 0(b)
> Can't exactly remember the byte order.
> The important point is there used to be a special case where transparent black was given a 32bit  pixel-value of 1 rather than 0.
>
> Cheers,
> Henry
>
> _______________________________________________
> Pharo-project mailing list
> [hidden email]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project

> do you have an idea how to fix that?
>
> On Jun 6, 2010, at 3:36 AM, Henrik Sperre Johansen wrote:
>
>> On 06.06.2010 03:18, Henrik Sperre Johansen wrote:
>>> On 05.06.2010 20:10, Oscar Nierstrasz wrote:
>>>> No, we have other tests that fail only when running all tests -- those that test finalization.
>>>>
>>>> The PNG tests fail reliably in 1.1.  Unfortunately we were unable to narrow down the cause.
>>>>
>>>> There appears to be an off by one error somewhere in encoding (or decoding) the Color black.
>>>>
>>>> - on
>>> Related to:
>>> http://code.google.com/p/pharo/issues/detail?id=2486
>>> perhaps?
>>> Essentially, 32 bit transparent black used to be 0(r) 0(g) (1)b (0)a instead of 0(r) 0(g) 0(b) 0(a).
>>> Which, was wrong. (And caused trouble when using transparent black as one of the colors in a bitblt operation)
>>>
>>> Cheers,
>>> Henry
>> or  (1)a 0(r) 0(g) (0)b instead of 0(a) 0(r) 0(g) 0(b) / (0)a 0(r) 0(g) (1)b instead of 0(a) 0(r) 0(g) 0(b)
>> Can't exactly remember the byte order.
>> The important point is there used to be a special case where transparent black was given a 32bit  pixel-value of 1 rather than 0.
>>
>> Cheers,
>> Henry
>>
>> _______________________________________________
>> Pharo-project mailing list
>> [hidden email]
>> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>
> _______________________________________________
> Pharo-project mailing list
> [hidden email]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project

> Recreate PNGReadWriterTest>>coloredFiles32, which was generated from the old pixel value of black.
>
> SLICE-Issue-2512-PNGTestsBasedOnFaultyBlack in Inbox.
>
> Cheers,
> Henry
>
> On 06.06.2010 20:04, Stéphane Ducasse wrote:
>> do you have an idea how to fix that?
>>
>> On Jun 6, 2010, at 3:36 AM, Henrik Sperre Johansen wrote:
>>
>>> On 06.06.2010 03:18, Henrik Sperre Johansen wrote:
>>>> On 05.06.2010 20:10, Oscar Nierstrasz wrote:
>>>>> No, we have other tests that fail only when running all tests -- those that test finalization.
>>>>>
>>>>> The PNG tests fail reliably in 1.1.  Unfortunately we were unable to narrow down the cause.
>>>>>
>>>>> There appears to be an off by one error somewhere in encoding (or decoding) the Color black.
>>>>>
>>>>> - on
>>>> Related to:
>>>> http://code.google.com/p/pharo/issues/detail?id=2486
>>>> perhaps?
>>>> Essentially, 32 bit transparent black used to be 0(r) 0(g) (1)b (0)a instead of 0(r) 0(g) 0(b) 0(a).
>>>> Which, was wrong. (And caused trouble when using transparent black as one of the colors in a bitblt operation)
>>>>
>>>> Cheers,
>>>> Henry
>>> or  (1)a 0(r) 0(g) (0)b instead of 0(a) 0(r) 0(g) 0(b) / (0)a 0(r) 0(g) (1)b instead of 0(a) 0(r) 0(g) 0(b)
>>> Can't exactly remember the byte order.
>>> The important point is there used to be a special case where transparent black was given a 32bit  pixel-value of 1 rather than 0.
>>>
>>> Cheers,
>>> Henry
>>>
>>> _______________________________________________
>>> Pharo-project mailing list
>>> [hidden email]
>>> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>>
>> _______________________________________________
>> Pharo-project mailing list
>> [hidden email]
>> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
>
>
> _______________________________________________
> Pharo-project mailing list
> [hidden email]
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project