On 08/17/2011 01:14 AM, Carsten Haitzler (The Rasterman) wrote:
> On Wed, 17 Aug 2011 02:27:30 +0300 Teodor Petrov<fuscated@...> said:
> the problem is related to the fact that the string length/size calculation
> thinks its "100" pixels wide, but when it actually renders the last char spills
> over to 102 or 103 pixels. it didnt calculate this at initial string geometry
> calc. the question is.. why? is there a bug in the geometry calc or render
> code that means they use different geometry when calculation character
> positions, or is the bounding box size given for the last char wrong? is it an
> issue in kerning caches ans chosing the wrong cache entry the 2nd time around?
> don't know - but it requires someone who reproduces the bug to sit down and
> throw some printf's into the query code in evas and the draw code per string
> and check the numbers patch per char etc. first. they very likely don't - then
> figuring out what differs and then why.
Thanks for the insight mate ;) I'll do some printf-ing and such @ a
latter date when I have less things todo. Heading to sleep now as it's
late here, but hopefully someday soon I can get some time for this as
it's quite annoying and apparently I am not the only one getting it, but
I can reproduce it here always.
>> On 08/16/2011 10:26 AM, Tom Hacohen wrote:
>>> Would you mind running the attached test.edj with edje_player?
>>> The red rectangle should be the size of the text object, so it's very
>>> easy to see if something bad is going on wrt to sizing.
>>> Also, you can decompile it and try different fonts on your system.
>>> Please let me know if you are getting good/bad results.
>>> Also, what e theme are you using?
>> I have this problem, too.
>> I'm running r62516 on gentoo linux ~amd64 (aka testing). XOrg 1.10.3.
>> Nvidia drivers 275.09.07.
>> I have no compositing and there is a warning/error dialog at the start
>> of E17 about it.
>> I've tried the test.edj and it worked correctly.
>> Except for some warnings:
>> ERR<27155>:edje edje_amalgamation.c:37695
>> edje_object_size_min_restricted_calc() file /tmp/test.edj, group main
>> has a non-fixed part '_tb1'. Adding 'fixed: 1 1;' to source EDC may
>> help. Continuing discarding faulty part.
>> I'm seeing another font issue. It happens when I type 3 characters in
>> the everything's search entry box.
>> The third character is rendered incorrectly. There is some kind of
>> tearing or double rendering.
>> Probably this problem is not related to the other.
>> Best regards,