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

Re: [Public WebGL] loop constants



Sorry my early replies confused you.  Top-of-tree chrome has the same
behavior as Firefox 4, that is, generating an error when non-constants
are used as loop bounds, which is what spec says.

Chrome 9 uses an outdated ANGLE that's apparently buggy on this issue.

Mo

On Sun, Jan 30, 2011 at 7:02 AM, John Davis <jdavis@pcprogramming.com> wrote:
> What's the final decision here?  Loop indexes are bounded by constants or
> non-constants?
>
> On Sat, Jan 29, 2011 at 11:23 PM, Benoit Jacob <bjacob@mozilla.com> wrote:
>>
>> I just updated Firefox's copy of ANGLE to the latest revision (550).
>>
>> So Sunday's nightly build will have the new behavior.
>>
>> Benoit
>>
>> ----- Original Message -----
>> > I just checked with ToT chromium build, and this demo works fine. So
>> > it's likely Firefox 4.0b10 uses an older version of ANGLE.
>> >
>> > My guess is in the latest Angle Shader translator gets smarter and
>> > realizes this variable is a constant (is this against the ES spec?).
>> > I can verify this on Monday when I get into office.
>> >
>> > Mo
>> >
>> > On Sat, Jan 29, 2011 at 8:13 PM, John Davis <jdavis@pcprogramming.com>
>> > wrote:
>> > > Hit this link with Firefox 4.0b10, it'll throw an error in the
>> > > shader
>> > > validator about the loop index being compared to a non-constant.
>> > > In Google Chrome 9 beta, the spinning cube with the multi-octave
>> > > texture
>> > > looks great.
>> > > http://www.pcprogramming.com
>> > > JD
>> > >
>> > > On Sat, Jan 29, 2011 at 3:14 PM, Vladimir Vukicevic
>> > > <vladimir@mozilla.com>
>> > > wrote:
>> > >>
>> > >> ANGLE is used in all cases to validate the shader first though,
>> > >> regardless
>> > >> of whether it's passed to the native GL driver or goes through
>> > >> ANGLE for d3d
>> > >> translation. John, can you post an example of a shader that you're
>> > >> talking
>> > >> about?
>> > >>
>> > >>    - Vlad
>> > >>
>> > >> ----- Original Message -----
>> > >> > On 01/29/2011 07:51 AM, John Davis wrote:
>> > >> > > In the fragment shader, when I run my stuff under Chrome
>> > >> > > variable
>> > >> > > loops work, while under FireFox they do not (I have to make
>> > >> > > them
>> > >> > > constants). If both implementations use Angle, this shouldn't
>> > >> > > be the
>> > >> > > case, right?
>> > >> > >
>> > >> > Firstly, variable shader loops (those where the termination
>> > >> > condition
>> > >> > isn't a constant) aren't supposed to work in WebGL - they are
>> > >> > disallowed
>> > >> > by the OpenGL-ES spec. However, it's possible that the checking
>> > >> > that
>> > >> > stops you from using them isn't working for some reason. It's not
>> > >> > necessarily the case that FireFox will use ANGLE...on some
>> > >> > systems it
>> > >> > uses the native OpenGL when Chrome chooses ANGLE...which could
>> > >> > certainly
>> > >> > explain the difference you're seeing between the two browsers.
>> > >> >
>> > >> > -- Steve
>> > >> >
>> > >> >
>> > >> > -----------------------------------------------------------
>> > >> > 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
>> > >> > -----------------------------------------------------------
>> > >> -----------------------------------------------------------
>> > >> 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
>> > >> -----------------------------------------------------------
>> > >>
>> > >>
>> > >
>> > >
>> >
>> > -----------------------------------------------------------
>> > 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
>> > -----------------------------------------------------------
>>
>> -----------------------------------------------------------
>> 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
>> -----------------------------------------------------------
>>
>>
>
>

-----------------------------------------------------------
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
-----------------------------------------------------------