Re: Hang in font panel with certain fonts

[Riccardo Mottola]

Hi,

Fred Kiefer wrote:
> Sorry, you only show these values for the first iteration. In a loop we need to 
> see whether these values change or not. Most likely the second iteration has 
> the exact same values. In that case we are looking at a rounding issue. It 
> seems like ascender plus descender are bigger than line_height, but when 
> assigning this sum to line_height this condition is still true.

It is not the first iteration, it is the hundreth or so, I leave running 
and then break and copy what I see. It is difficult to know in the 
console flow "when" the loop begins amidst all the printouts.

Ascender + Descender differ from line height here only by 0,00001

this is after a couple of seconds:

2015-01-19 00:19:35.047 Ink[322] 2-WantLineHeight a:10.285714, 
d:2.571429, y:0.000000 | lineh: 12.857142, maxl: 0.000000
2015-01-19 00:19:35.050 Ink[322] f_ascender: 10.285714

this is after a couple of minutes

2015-01-19 00:26:04.893 Ink[322] 2-WantLineHeight a:10.285714, 
d:2.571429, y:0.000000 | lineh: 12.857142, maxl: 0.000000
2015-01-19 00:26:04.896 Ink[322] f_ascender: 10.285714

it is exactly the same

> Could you please first verify whether this is true and next find out if your 
> computer uses float or double for CGFloat? The next step might then be to get 
> David to look at the assembler code the macro gets converted into.
How do I know what CGFloat is? being 32bit I suppose float, but where do 
we configure that?

I added this in the Macro:

    if (__new_height > line_height) \
      { \
        line_height = __new_height; \
        NSLog(@"lh = %f, h = %f, gt? %d", line_height, h, line_height > 
h); \
        COMPUTE_BASELINE; \
        goto restart; \
      } \

this is the output (other log removed):
2015-01-19 00:39:12.193 Ink[1887] lh = 12.857143, h = 12.857143, gt? 0


Riccardo