rearrange code

rebase

@bowonryu I think safest way is to search for the last
"font=" and "font_size="
Then replace the string in front of it (until you reach white space), with new value.

rebase

rebase

Can you take a look at D11990 ? I think it is good Idea to have this new function.

@raster multiplication is normally faster than division, So this should enhance the speed

remove len_2

I think this makes sense in term of division vs multiplication

exclude sin/cos changes

add FIXME comment

default behaviour is unlimited texture size

free cache on evas_font_texture_cache_set call (if needed), and change unlimited cache to be negative value instead of 0

change long to int for public APIs

this seems fine to me, as long it pass test for others (I think it will pass since it is not dependent on emoji font anymore)

This seems fine to me

I think changing this line to replace last occurrence instead of first will resolve the issue

Does this mean passing struct of rect instead of 4 int values ?

In D11962#228200, @bowonryu wrote:

If there is more than one font or font_size in the textblock style, the last tag is replaced.
Because if there are multiple tags, then tag->font, tag->font_size in code is the last tag.
As you say, textblock style may have multiple font or font_size, however only the last tag is valid.

What if the font style contains more than one font or font_size with same value ?
I know it seems strange but it is valid (could be done with pushed user styles or user made more than one tag).

In D11949#227998, @bu5hm4n wrote:

The code *is* the same, Add -O3 and you will just see that the code lines are not mapped to anything in asm. Even on the wegpage. The next example also gets eliminated the cos and sin calls.

In D11949#227994, @bu5hm4n wrote:

Don't look at the times, take a look at objdump -S a.out the outputted asm is equivalent. Also take a look where the calls to clock are located. The time advantage you see is *not* coming from the different call. infact, the sin and cos calls are stripped out completly. As a mixture of inlining, dead code removal, and const folding removed it.

@raster
First : this is very small impact.
Second : you are talking only about specific compiler(s) , or maybe most of them, but this is not general thing everywhere.

@raster lets look at the following code:

#include<math.h>
#include<time.h>
#include<stdio.h>
#include<stdlib.h>

In T8737#152896, @raster wrote:

can we not cal it texture_cache? as this also applies to software also not long but int will do like the other caches. if people need > 2gb of cache for this... they have problems :) so maybe evas_font_data_cache_set/get(). ?

If we are talking about GCC and its optimization, then the whole thing does not give much enhancement.

@bu5hm4n I am not sure I fully understand your comment, I use M_SQRT1_2 as its faster and readable (IMO).

use sincos instead of two calls for sin and cos

change 1/M_PI_2 to M_2_PI

In D11947#227828, @bu5hm4n wrote:

I dont mind if we land this or not. But this is not really impacting anything, const folding covered the cases i checked, and if not, then the static double assignment will fail...

In T8742#152883, @smohanty wrote:

@ali ,
I didn't know evas_object_textblock_style_set() only keeps the reference to the raw pointer.

Could you please change the testapp to fix the issue .

thanks

@zmike can you please confirm if latest update fix the test failing

Can we close this patch please, (unless there are some comments)

I think we should change the part that set visually_deleted to FALSE will remove from line items

This is an issue, D11945 should fix it, and release leaked data

I think this is leak in the demo it self, where it does not free style that was created.