Re: C file recoginzed as image file

[Stephen Leake]

"Juanma Barranquero" <address@hidden> writes:

> On 1/9/07, Stephen Leake <address@hidden> wrote:
>
>> "if the file extension does not match the contents, it is more likely
>> that this is a virus attack"
>
> "More likely" than what? More likely than the alternative: that it is
> not a virus attack?

Yes.

Obviously, this only makes sense in a weighting scheme, where "more
likely" is a useful concept.

In the current code, the choices are all binary, so this kind of
heuristic isn't much help

> I have mislabeled images in my hard disk: it's not unusual in my
> experience to find an image in format A, with an extension suggesting
> that it is in format B. So far, none of them have been virus attacks.

Ok.

> Are you proposing also that we reject 

no

> (or warn about) 

yes, but allow users to easily disable the warning.

> a .PNG file disguised as a .JPG, for example? We don't have a
> png-mode and a jpeg-mode; the auto-detection and image opening
> machinery just return image-mode. Should we redesign it to take that
> into account? 

Perhaps. Certainly not now. It depends on how much effort we want to
devote to preventing virus attacks, versus all the other things we
need to do.

-- 
-- Stephe