On Tue, 11 Oct 2011 08:40:33 +0200 (CEST) Vincent Torri <vto...@univ-evry.fr>
said:
>
>
> On Tue, 11 Oct 2011, Carsten Haitzler (The Rasterman) wrote:
>
> > On Tue, 11 Oct 2011 08:14:47 +0200 (CEST) Vincent Torri
> > <vto...@univ-evry.fr> said:
> >
> >>
> >>
> >> On Mon, 10 Oct 2011, Enlightenment SVN wrote:
> >>
> >>> Log:
> >>> use lround() for map coord rounding to avoid silly things like
> >>> 15.999999999999999998 rounding down to 15... whihc leads to
> >>> sometimes... odd off-by-1 expected results.
> >>
> >> lround is C99 compliant and does not exist on Windows.
> >>
> >> why not using the macro
> >>
> >> #define round(x) (x<0?ceil((x)-0.5):floor((x)+0.5))
> >
> > because 1. i dont have a manual page for what is NOT supported in windows's
> > dev envs (that's why we have you!).
>
> in google : "msdn the_function"
>
> Also, did you read the man page of lround ?
i did.
...
long long int llroundl(long double x);
Link with -lm.
Feature Test Macro Requirements for glibc (see feature_test_macros(7)):
All functions shown above:
_XOPEN_SOURCE >= 600 || _ISOC99_SOURCE ||
_POSIX_C_SOURCE >= 200112L;
or cc -std=c99
...
CONFORMING TO
C99, POSIX.1-2001.
...
nothing in the manual page on ubuntu requires any special compile flags other
than -lm (as per usual) for linking. the manual differs from the one you quote
below. note that the manula i have seems more complete. as such the standard
for efl is c99 as an assumed base.
> See
>
> http://linux.die.net/man/3/lround
>
> did you read the line "Compile with -std=c99" ? Is that flag passed to the
> compiler ? Note that i added an m4 macro that checks if a flags is
> supported by the compiler
>
> 2. floor and ceil return doubles, not
> > integers. lround returns a long int (long).
>
> so (long)ceil(15.999999999999999) does not return 16 ?
correct. in some cases where 15.9999999999999 cannot in fp be represented as
a value >= 16.0, this it gets slightly rounded down, and thus ends up being 15
when dropped down to an integer. that is the whole crux of the problem.
> > why not do it that way? less efficient than a single lround call. and
> > technically incorrect. i'm literally trying to solve minuscule rounding
> > issues here and returning a rounded double could still return an inaccurate
> > (by a very tiny margin) double that will still get rounded the wrong way. :
> > ( need a call that rounds and returns an integer type (int, long, etc.) not
> > a double or float.
>
> The, at least, check the availability in configure.ac
so that means non-c99 platforms will have rendering bugs. there has to be a
better way.
> then
>
> #ifndef HAVE_LROUND
> # define lround ****
> #endif
>
> Vincent
>
> ------------------------------------------------------------------------------
> All the data continuously generated in your IT infrastructure contains a
> definitive record of customers, application performance, security
> threats, fraudulent activity and more. Splunk takes this data and makes
> sense of it. Business sense. IT sense. Common sense.
> http://p.sf.net/sfu/splunk-d2d-oct
> _______________________________________________
> enlightenment-devel mailing list
> enlightenment-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
>
--
------------- Codito, ergo sum - "I code, therefore I am" --------------
The Rasterman (Carsten Haitzler) ras...@rasterman.com
------------------------------------------------------------------------------
All the data continuously generated in your IT infrastructure contains a
definitive record of customers, application performance, security
threats, fraudulent activity and more. Splunk takes this data and makes
sense of it. Business sense. IT sense. Common sense.
http://p.sf.net/sfu/splunk-d2d-oct
_______________________________________________
enlightenment-devel mailing list
enlightenment-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
