Details
-
Type:
Improvement
-
Status: Closed
-
Priority:
Major
-
Resolution: Fixed
-
Affects Version/s: 3.1
-
Fix Version/s: 3.1
-
Component/s: Core/Parsing
-
Labels:None
-
Environment:any
Description
The first big offender was GraphicsEnvironment.getLocalGraphicsEnvironment().getAllFonts() was being called for every new instance of a ofont/Font object. This system call is pretty expensive.
The next big offender was in related to a sequence of awt calls to get a glphys x and y advance values. I change the code to cache the metric information and only do the full calculation when we need to.
Two very simple optimizations that increased the parsing speed significantly for the OS release.
Activity
- All
- Comments
- History
- Activity
- Remote Attachments
- Subversion
There were a few regression in the the image comparor that showed incorrect glyph width calculations for layout. I had logic error when calculating a font point size change which failed to clear the echarAdvanceCache map. Silly mistake.
Qa found a regression with font width which is related to this bug.
It turns out the the deriveFont() call where being used for different font instances and as a resulted the cached widths weren't always correct. Update code slightly to recalculate width on the fly but still skip the heavy system calls. I aslo added code to clear the cache when deriveFont is called with metrics changing data.
Found an issue in where the glyph width was incorrectly being calculated after the optimization.
Checked in source, closing.
ICEpdf 3.1.0 has been released, closing issues.