Issue with BitBltSimulation

Hi All,

We are currently trying to use the BitBltSimulation (VMMaker.oscog-topa.2418) instead of the native BitBltPlugin. We ran into problems when trying to access a font form via e.g. StrikFont class >> #dejaVuSansBold9Form. We traced the issue down to PNGReadWriter >> copyPixelRGB: at which point the correctly decoded image data is copied via BitBlt into the target form. This blt operation appears to be faulty for the BitBltSimulation.

When executing

StrikeFont dejaVuSansBold9Form asMorph openInWorld.

using the native BitBlt we generated the correct form:

As soon as we change to the BitBltSimulation (by e.g. changing the message copyBits in PNGReadWriter >> copyPixelRGB: to copyBitsSimulated), the output changes to a completely transparent form. If we then set the alpha channel for our output image to 255 we see the following:

It seems to us that the correct image data is somehow stretched and offset during the blting.

-- Stephan

> On 21-06-2018, at 6:32 AM, Stephan Lutz <[hidden email]> wrote:
>
> It seems to us that the correct image data is somehow stretched and offset during the blting.

That looks like the effect one gets when the pixel depth is wrong, or the target word width gets miscalculated etc. Try checking the various width & height values that get calculated, the bits per pixel, stuff like that. Remember also that we have to mangle pixels being read from outside files because our pixel value 0 is transparent and not black!

tim
--
tim Rowledge; [hidden email]; http://www.rowledge.org/tim
Strange OpCodes: IPL: Invent Program Lines