[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Public WebGL] readPixels overloads?

On 4/20/12 11:22 PM, Glenn Maynard wrote:
    Thinking about readPixels some more, would it make sense to add an
    overload like this to the spec:

      void readPixels(GLint x, GLint y, GLsizei width, GLsizei height,
                      ArrayBufferView? pixels)

    which assumes format=RGBA and type=UNSIGNED_BYTE?  If one really
    wanted, one could even make it take a Uint8Array; I think the
    current behavior of allowing passing in the wrong array types but
    "silently" not writing data to them is sorta hostile as APIs go.

The problem wouldn't exist if functions threw exceptions for invalid
enum arguments (and other things which are almost always programming
errors and which can be detected without flushing the render queue), but
failing that it doesn't seem worth sidestepping it for just a single API
call.  Other functions do the same thing if you pass an invalid enum (do
nothing except generate an error).

The main problem I'm trying to solve here is not the "do nothing" behavior but "this function requires two arguments whose values must always be the same constant, which seems like bad API".

    This would not break existing code (or future code if more
    modes/formats ever get added), since the old overload would still be
    there, but would allow simpler code to be written in the meantime.

(Omitting two parameters here isn't much of a simplification--this isn't
a function that's used a whole lot; rarely more than once or twice in an
application, if that.)

<shrug>. Just pointing out things that would make me, as someone not intimately familiar with WebGL, go "WTF?" if I had to write code to the API.


You are currently subscribed to public_webgl@khronos.org.
To unsubscribe, send an email to majordomo@khronos.org with
the following command in the body of your email:
unsubscribe public_webgl