This applies the font size inflation to reflow and painting of text
frames. However, it does not (by design) apply to intrinsic width
computation, since the inflation is itself a function of the containers
width, which can depend on the intrinsic width.
This change is sufficient because the places that set mHResize to true
other than InitResizeFlags and nsFrame::BoxReflow aren't a problem
because they're in table code whose goal is to force the reflow to
propagate down to the cell, and once we reach the cell we'll hit the
code we've added here.
This implements computation of the font size inflation factor for a
given frame. Since Fennec does layout using a fake viewport whose width
represents a typical viewport width on the desktop and then allows users
to pan and zoom, fonts are not always readable even when zoomed. The
goal of this font size inflation is to ensure that when a block of text
is zoomed to fill the width of the device, the fonts are large enough to
read. We do this by increasing the font sizes in the page. Since this
increase is a function of the width of the text's container, the
inflation must be performed (in later patches in this series) after
style data computation and after intrinsic width computation.
The font size inflation factor does not vary *within* a block.
Since sync uses a whitelist (the services.sync.prefs.sync.* prefs) for
preferences (i.e., preferences are not synced by default), this patch
does not make any changes relating to sync, since we do not want the
inflation preferences synced across devices (since preferred settings
are likely to be device-specific).
This property is analogous to the -webkit-text-size-adjust property (and
*maybe* also the -ms-text-size-adjust property). It allows pages to opt
out of text size adjustments performed on mobile devices by specifying
-moz-text-size-adjust: none.
The version of llvm-gcc included with Apple's XCode 4.2
toolchain can't compile the inline assembly in the libtheora
library. This is a regression from the 4.1 release.
Check for this version and disable assembly if found.
Also update the check for clang < v2.9 from bug 627981,
which had similar problems, to set the same flag.
Currently we only run this on idle when allocating memory. There
are plans in place to also run these when clicking the Minimize
Memory button in about:memory, and when we are under memory
pressure. This change causes GC_SHRINK to also run when doing so
will result in "lots" of pages being decommitted to the OS.
