Re: OLEAUT32: fix detection of GIF palette index >= 128 for transparency

[Andreas Mohr]

Hi,

On Tue, Mar 29, 2005 at 12:43:38PM -0500, Alex Villaci­s Lasso wrote:
> I traced this problem to a one-line bug in the GIF loading code. The 
> code is supposed to initialize a transparency flag to -1 when no 
> transparency is used, or else the palette index of the transparency 
> color. The sample image I was using indicates that the transparency 
> index is 255. Due to an unintended sign extension, the transparency 
> index (held in a Byte array of basic type "signed char") is 
> sign-extended to the variable "transparency" of basic type "signed int". 
> This causes the flag to be assigned the value -1, which happens to be 
> the same value used as a no-transparency indicator. So transparency was 
>                                                      ....^^^^^^^^^^^^

Heh, 3 times "the transparency" in exactly the same line position, 4 times
"transparency" in exactly the same position - what did you smoke?? ;-)

(or worse, what did *I* smoke to recognize this?? :-)

OK, I promise I'll stop that useless fun email here.

Anyway, thanks for all those nice patches,

Andreas Mohr

OLEAUT32: fix detection of GIF palette index >= 128 for transparency

[Marcelo Duarte]

Hi,
Andreas Mohr escreveu:
> Hi,
>
> On Tue, Mar 29, 2005 at 12:43:38PM -0500, Alex Villaci­s Lasso wrote:
> > I traced this problem to a one-line bug in the GIF loading code. The 
> > code is supposed to initialize a transparency flag to -1 when no 
> > transparency is used, or else the palette index of the transparency 
> > color. The sample image I was using indicates that the transparency 
> > index is 255. Due to an unintended sign extension, the transparency 
> > index (held in a Byte array of basic type "signed char") is 
> > sign-extended to the variable "transparency" of basic type "signed int". 
> > This causes the flag to be assigned the value -1, which happens to be 
> > the same value used as a no-transparency indicator. So transparency was
>
> ....^^^^^^^^^^^^
>
> Heh, 3 times "the transparency" in exactly the same line position, 4 times
> "transparency" in exactly the same position - what did you smoke?? ;-)
>
> (or worse, what did *I* smoke to recognize this?? :-)

You is funny.... I am thinking what you smoked.
What I see is that Alex didn't cut the text, but your email client: yes!
> OK, I promise I'll stop that useless fun email here.
>
> Anyway, thanks for all those nice patches,
>
> Andreas Mohr

;)