issue was limited to osx and windows
before this, if you create a canvas and get its context, the fonts would
freeze - you cannot add any fonts after getContext is called once, even
if you set up a new canvas. i fixed this by refreshing the Pango FontMap
(the component that does font lookups)
after you create a canvas+context, that canvas+context will not be able
to use new fonts, but that is less of a problem
this is implemented differently depending on the operating system. for
MacOS, CTFontManagerRegisterFontsForURL is the call to add a temporary
font. for Linux we use FontConfig, and for Windows use
AddFontResourceEx. all have been tested successfully
updated README and examples/font.js
* this commit gets rid of the HAVE_PANGO pragma and HAVE_FREETYPE,
preferring the Pango sections of HAVE_PANGO. using Cairo methods for
fonts is discouraged (http://cairographics.org/FAQ/#using_pango),
and freetype was only ever needed for cairo font rendering, so that
code was running in the Pango build for no reason at all.
* it also gets rid of context2d's `addFont`, which was never compatible
with the Pango build. `FontFace` is completely removed since it was
never used by the Pango build either, and instead of parsing the fonts
with FreeType we will instead prefer to let the OS do that in
succeeding commits
* the Windows build has been updated to work with Pango, which it did
not before
TL;DR - Pango is now required and and as of this commit custom fonts are
removed
Remove the undocumented 3rd argument of the emitters (bytes left in buffer).
Add a column to the browser tests that displays JPEGs.
Revise how buffers are allocated.
From "[WHATWG HTML Living Standard]
(https://html.spec.whatwg.org/multipage/indices.html#event-load)", the definition of 'onload':
> Fired at the Window when the document has finished loading; fired at an
element containing a resource (e.g. img, embed) when its resource has finished
loading
To adhere to that specification both handlers should be called as many times as Image loads.
From Google Chrome's Developer Tools:
```js
> function handler(evt) { console.log('img handler: %s', evt.type); }
< undefined
> var img = document.createElement('img')
< undefined
> img.onload = handler
< handler(evt)
> img.onerror = handler
< handler(evt)
> img.src = 'https://www.google.com/images/errors/logo_sm_2.png'; true
< true
< img handler: load
> img.src = 'https://www.google.com/images/errors/logo_sm_2.png'; true
< true
< img handler: load
> img.src = 'https://example.com/404.png'; true
< true
< img handler: error
> img.src = 'https://example.com/404.png'; true
< true
< img handler: error
```
Benchmark:
var canvas = new Canvas(300, 600);
var ctx = canvas.getContext("2d");
// any manipulation of canvas/ctx here.
var data = ctx.getImageData(0,0,300,600);
// time 1000x:
ctx.putImageData(data, 0, 0);
Added some argument testing/manipulation to match WebKit/Moz behaviors.
Additionally benchmarked:
* branching on `if (a == 0 || a == 255)`, as it is
* not branching (always doing the alpha calculation)
* Mozilla's implementation found here: https://dxr.mozilla.org/mozilla-central/source/dom/canvas/CanvasRenderingContext2D.cpp#5083
Mozilla's is insignificantly faster (p=0.17) :) so left it as-is.
Caleb Hearon
Adam Hooper
rhysd
Benjamin Byholm
wsw
arusakov
Linus Unnebäck
Jan Stola
Raul Ochoa
Zach Bjornson
Radosław Eichler
Hussein Jafferjee
Jesper Ek
Michaël Zasso
Nathan Rajlich
Кузнецов Дмитрий
kibao
Alexander Parshin
Stephen Remde
Josip Janžić
Joe Freeman
