On Sat, Jun 7, 2008 at 4:46 PM, Jose Gonzalez <[EMAIL PROTECTED]> wrote:
> Cedric wrote:
>>>> Well, the semantics of the image-size-set function has always
>>>> been that if there is no data or file for the image, it will create
>>>> data of the requested size, and if there is data (possibly from a
>>>> file), it will create new data of the requested size and copy as much
>>>> as possible of the current data to the new one and then set that new
>>>> one as the current data.
>>>> So, for the test program there, after the inital rendering
>>>> (of a blue area), when the resize is called the image would have
>>>> new data (not the provided pixels) one pixel less in width, so the
>>>> subsequent ecore induced rendering would still show the same result
>>>> (even though there's no further call to 'set data' as the callback
>>>> won't get triggered since 'dirty pixels' isn't called again, and
>>>> it's been unset after the first rendering).
>>>>
>>>> This would seem to show some new problem with the 'image-resize'
>>>> function when external data has been set before on the image.. and
>>>> indeed if I get rid of the pixels callback and simply set the data
>>>> (after the initial image size setting), then the problem is still
>>>> there even without the first call to evas render.. ie. the second
>>>> call to image-size-set seems to be the culprit.
>>>
>>> Took a quick look at the software_generic engine, and in the file
>>> evas_engine.c, we see the following implementations of say:
>>>
>>> static void *
>>> eng_image_size_set(void *data, void *image, int w, int h)
>>> {
>>> RGBA_Image *im;
>>>
>>> im = image;
>>> return evas_cache_image_size_set(image, w, h);
>>> }
>>>
>>> static void *
>>> eng_image_dirty_region(void *data, void *image, int x, int y, int w, int
>>> h)
>>> {
>>> RGBA_Image* im = image;
>>>
>>> if (!image) return NULL;
>>> im = (RGBA_Image *) evas_cache_image_dirty(&im->cache_entry, x, y, w,
>>> h);
>>> return im;
>>> }
>>>
>>> We may contrast these with the signatures of the called 'cache'
>>> functions, in the evas_cache_image.c file:
>>>
>>> EAPI Image_Entry *
>>> evas_cache_image_size_set(Image_Entry *im, int w, int h);
>>>
>>> EAPI Image_Entry *EAPI Image_Entry *
>>> evas_cache_image_dirty(Image_Entry *im, int x, int y, int w, int h)
>>>
>>>
>>> So, it may be that something's not quite correctly called here.
>>> It's also possible that the actual implementation of the function
>>> "evas_cache_image_size_set" isn't doing 'the right thing', but haven't
>>> had a chance to look... Cedric?
>>>
>>
>> No this is correct. But eng_image_dirty_region is not called in the
>> second case. If I force the dirty_pixel to 1, it call
>> eng_image_dirty_region and every thing goes ok. So perhaps in case we
>> have func.get_pixels set, we should on resize set dirty_pixel to 1. I
>> think this should sove the issue, but I am not sure it's really the
>> correct fix.
>>
>
> Something's not correct with the image-size-set function,
> it should be taking the externally set pixel data, and copying
> it to newly created data and setting that as the image data.
> Forget about the pixels-get function in that example, just
> comment it out and instead add a call to the image-data-set func
> instead... I've included a simplified version of their 'error-demo'
> below which does this. It should work correctly but it doesn't.
> :(
Hum, if I add in evas_common_rgba_image_size_set (evas_image_data.c) a
call to evas_common_blit_rectangle, it fix the problem. If you think
it's the right solution I will commit the change.
/***/
int
evas_common_rgba_image_size_set(Image_Entry *ie_dst, const Image_Entry
*ie_im, int w, int h)
{
RGBA_Image *dst = (RGBA_Image *) ie_dst;
RGBA_Image *im = (RGBA_Image *) ie_im;
if ((im->cache_entry.space == EVAS_COLORSPACE_YCBCR422P601_PL) ||
(im->cache_entry.space == EVAS_COLORSPACE_YCBCR422P709_PL))
w &= ~0x1;
dst->flags = im->flags;
dst->cs.no_free = 0;
if ((im->cache_entry.space == EVAS_COLORSPACE_YCBCR422P601_PL) ||
(im->cache_entry.space == EVAS_COLORSPACE_YCBCR422P709_PL))
dst->cs.data = calloc(1, dst->cache_entry.h * sizeof(unsigned char *) * 2);
evas_common_image_colorspace_dirty(dst);
evas_common_image_colorspace_dirty(im);
evas_common_blit_rectangle(im, dst, 0, 0, w, h, 0, 0);
evas_common_cpu_end_opt();
return 0;
}
/***/
--
Cedric BAIL
-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://sourceforge.net/services/buy/index.php
_______________________________________________
enlightenment-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
