Chrome ticket: https://bugs.chromium.org/p/chromium/issues/detail?id=719866

Am Dienstag, 9. Mai 2017 10:00:13 UTC+2 schrieb Floh:
>
> Ok, please ignore the title, it's not in emscripten, it's the latest 
> Chrome Canary.
>
> I was confused because the problem only happens when compiling with -O2 or 
> -O3, not with -O0 or -O1.
>
> Chrome stable and Firefox don't have the problem.
>
> You can check yourself in this demo: 
> http://floooh.github.io/oryol/asmjs/DebugText.html, in Chrome Canary, the 
> characters are very small and in the top-left corner.
>
> I'll write a ticket on the Chrome bug tracker.
>
> Am Dienstag, 9. Mai 2017 00:06:10 UTC+2 schrieb Floh:
>>
>> Interestingly, the problem does *not* happen when compiling for 
>> WebAssembly, it only happens in asm.js. The only difference is "-s WASM=1" 
>> and "-s BINARYEN_METHOD="native-wasm"" in the linker stage.
>>
>> Am Montag, 8. Mai 2017 23:47:40 UTC+2 schrieb Floh:
>>>
>>> Hi, 
>>>
>>> I just noticed a very strange regression after I upgraded to the latest 
>>> incoming last Friday:
>>>
>>> Basically this fairly old code produces the wrong values for GlyphSize 
>>> now:
>>>
>>>
>>> https://github.com/floooh/oryol/blob/master/code/Modules/Dbg/text/debugTextRenderer.cc#L133
>>>
>>> It's not the glm::vec2 class or any alignment issue, the issue also 
>>> happens when rewriting like this:
>>>
>>> const float w = 8.0f / Gfx::PassAttrs().FramebufferWidth
>>> const float glyphWidth = w * 4.0f;
>>>
>>> (FramebufferWidth is an int, usually 800, the 4.0f factor is from 2.0f * 
>>> textScale (textScale being 2.0f)). The resulting glyphWidth is too small 
>>> (0.0025f instead of 0.04f).
>>>
>>> Rewriting the expression to (32.0f / Gfx::PassAttrs().FramebufferWidth) 
>>> works correctly.
>>>
>>> Compiling a native version in Xcode produces correct code. It must be a 
>>> very recent emscripten regression (or clang 4.0?).
>>>
>>> Anybody seen similar problems after updating incoming?
>>>
>>> It's quite late now, I'll try to find out more tomorrow (do a fresh 
>>> install from scratch, write isolated test case, do a bisect, etc...).
>>>
>>> Cheers,
>>> -Floh.
>>>
>>>

-- 
You received this message because you are subscribed to the Google Groups 
"emscripten-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to