Thread: Album Art
View Single Post
Old 25th January 2012, 17:53   #19
DrO
 
Join Date: Sep 2003
Posts: 27,873
i know what you're doing and it's the same as the people sending things to the list about localisation issue.

i can no longer properly attempt to look at the 'A' issue since i don't have a true XP install anymore and doing it via a limited XP VM is not something i really want to do. i did try to look into it as well when i had a real XP machine and was _never_ able to reproduce it.

Quote:
Originally Posted by Nitorami View Post
On the assembler: I thought MD was written after 2000 ? Who would still use assembler ? It may be worth pointing out that the issue was, to my recollection, NOT introduced with the ns-eel language extensions a while ago. Unlike loops, megabufs etc., the pow() function existed in older versions already and it always had the error.
most of the plug-in is c/c++ but as ns-eel uses assembler to do certain aspects in an attempt to do things faster or something like that - it's just how it's implemented. things still are written in assembler if trying to achieve a certain type of performance, etc so just because of where we are, doesn't mean assembler is not relevant - it's just i'm somewhat rusty on such things and i'd prefer to fix the implementation used without having to try to change everything else in ns-eel2 (as that's far far outside of the remit of what i can warrant doing off my own back considering the list of things i actually need to work on).

it does seem to be it's only with < 1 values where things are being incorrectly rounded (have spent more time messing around looking at it instead of working on what i should really have been doing) though i need to properly see what is / isn't being passed through the internal function to work out the cause of the issue (which my gut makes me think it's how rounding of the values is handled).

-daz
DrO is offline   Reply With Quote