Re: [E-devel] ecore: request to drop printf warning

[Dave R. <cl...@jo...>]

It's a known issue with Darwin, which MacOS is part of. They haven't had clock_gettime support for 6+ years.

There are alternative time calls that work on Darwin. but the fallback in ecore seems to work fine. There are some good discussions on the net, I can post some ideas for monotonic clocks if interested. 

I can try adding the EINA flag to suppress the warnings, that sounds like the best option for now.

But I wonder does anyone benefit from this printf warning spewing frequently. It seems to work fine using the fallback.


On Apr 8, 2011, at 6:06 PM, Carsten Haitzler (The Rasterman) wrote:

> On Fri, 8 Apr 2011 21:36:39 -0300 Gustavo Sverzut Barbieri
> <bar...@pr...> said:
> 
>> On Fri, Apr 8, 2011 at 8:40 PM, Dave Ray <cl...@jo...> wrote:
>>> On my OS ecore runs fine, but spews a warning frequently.
>>> 
>>> CRI<12490>:ecore ecore_time.c:170 _ecore_time_init() Platform does not
>>> support clock_gettime. Fallback to unix time.
>>> 
>>> Everything that uses ecore spews it. Fills up my logs.
>>> 
>>> Is this printf necessary?
>> 
>> It's not a printf(), but eina_log and you van disable it with
>> EINA_LOG_LEVELS=ecore:-1
>> 
>> What platform is yours? The correct fix would be to add proper
>> monotonic clock to it... this may result in skews and problems during
>> timezone changes.
> 
> not just timezone - every time the clock is changed - ie u set the time (ntp
> adjusts clock skew, etc. etc) depending on timezone setup and so on. i would
> agree with gustavo - your Os sounds pretty poor if it has no monotonic clock.
> if it does and it simply has decided to not comform to posix (As clock_gettime
> is posix.1-2001) then it's just wanting to be different for the sake of being
> different. if it is a problem with our detection of the call and it does exist,
> then please let us know what it requires to detect it etc. (see configure.ac
> for ecore - we check libc and if not we check librt) :)
> 
> -- 
> ------------- Codito, ergo sum - "I code, therefore I am" --------------
> The Rasterman (Carsten Haitzler)    ra...@ra...
>