[Olsr-dev] [PATCH RFC v1 0/4] Smart Gateway dynamic bandwidth

Hans-Christoph Steiner (spam-protected)
Fri May 11 01:24:10 CEST 2012


Mac OS X claims that defining _POSIX_C_SOURCE will give you:

	time_t		st_mtime;	/* [XSI] Last data modification time */
	long		st_mtimensec;	/* last data modification nsec */

It seems that NetBSD agrees:
http://www.daemon-systems.org/man/stat.2.html

So it would be easy to add *BSD support it seems.

.hc


On May 10, 2012, at 5:28 PM, Ferry Huberts wrote:

> that's not the point
> mtime is in _seconds_ accuracy.
> a file can very easily change within a second...
> 
> On 10-05-12 23:03, Hans-Christoph Steiner wrote:
>> 
>> Given that it is highly unlikely that a file would change on a time scale less than 1ms, I think its safe to use the generic versions, at least on other platforms, no?  I think you need a special Linux kernel to get thread/process time slices with smaller than 1-2ms anyhow.
>> 
>> .hc
>> 
>> On May 10, 2012, at 3:49 PM, Ferry Huberts wrote:
>> 
>>> 
>>> It's not a typo.
>>> It appears that this is a linux extension: it provides nanosecond accuracy to stat system calls.
>>> 
>>> On 10-05-12 21:05, Hans-Christoph Steiner wrote:
>>>> 
>>>> Also, it looks like the Mac OS X build issue is actually caused by a typo in the code:
>>>> 
>>>> src/gatewaySpeedFile.c: In function ‘readGatewaySpeedFile’:
>>>> src/gatewaySpeedFile.c:124: error: ‘struct stat’ has no member named ‘st_mtim’
>>>> src/gatewaySpeedFile.c:134: error: ‘struct stat’ has no member named ‘st_mtim’
>>>> src/gatewaySpeedFile.c:134: error: ‘struct stat’ has no member named ‘st_mtim’
>>>> 
>>>> I believe the struct element is called st_mtime not st_mtim.  Both Ubuntu and Mac OS X have stat.st_mtime, but I find no reference to stat.st_mtim.
>>>> 
>>>> .hc
>>>> 
>>>> On May 10, 2012, at 2:58 PM, Henning Rogge wrote:
>>>> 
>>>>> At the moment the whole smart gateway system is linux only too... ;)
>>>>> 
>>>>> Henning
>>>>> 
>>>>> On Thu, May 10, 2012 at 8:54 PM, Ferry Huberts<(spam-protected)>   wrote:
>>>>>> 
>>>>>> 
>>>>>> ok, then.
>>>>>> Tomorrow I'll refactor it into a plugin but I'll make it linux only.
>>>>>> It people want it on other platforms as well, they can make it build
>>>>>> themselves ;-)
>>>>>> 
>>>>>> On 10-05-12 20:46, Henning Rogge wrote:
>>>>>>> 
>>>>>>> As they can with plugins... its not that it needs to be a linux-only
>>>>>>> plugin.
>>>>>>> 
>>>>>>> Henning
>>>>>>> 
>>>>>>> On Thu, May 10, 2012 at 8:25 PM, Ferry Huberts<(spam-protected)>    wrote:
>>>>>>>> 
>>>>>>>> 
>>>>>>>> 
>>>>>>>> On 10-05-12 20:22, Henning Rogge wrote:
>>>>>>>>> 
>>>>>>>>> 
>>>>>>>>> On Thu, May 10, 2012 at 7:17 PM, Ferry Huberts<(spam-protected)>
>>>>>>>>>  wrote:
>>>>>>>>>> 
>>>>>>>>>> 
>>>>>>>>>> Thinking about this...
>>>>>>>>>> 
>>>>>>>>>> I added this to the core, but what do you guys think of making this a
>>>>>>>>>> plugin? I think that should be possible, right Henning?
>>>>>>>>> 
>>>>>>>>> 
>>>>>>>>> Might work as a plugin I think, the only think it does in the core is
>>>>>>>>> set the link up/down speed. That should be easy to do from a plugin
>>>>>>>>> during runtime.
>>>>>>>>> 
>>>>>>>>>> However, I'm _not_ saying I want to make it a plugin because I think
>>>>>>>>>> this
>>>>>>>>>> should be part of the core, but a plugin might mitigate some of the
>>>>>>>>>> objections.
>>>>>>>>> 
>>>>>>>>> 
>>>>>>>>> 
>>>>>>>>> Whats the advantage to put it into the core?
>>>>>>>> 
>>>>>>>> 
>>>>>>>> 
>>>>>>>> then all platforms can benefit from it
>>>>>>>> 
>>>>>>>>> 
>>>>>>>>> Henning
>>>>>>>>> 
>>>>>>>> 
>>>>>>>> --
>>>>>>>> Ferry Huberts
>>>>>>> 
>>>>>>> 
>>>>>>> 
>>>>>>> 
>>>>>> 
>>>>>> --
>>>>>> Ferry Huberts
>>>>> 
>>>>> 
>>>>> 
>>>>> --
>>>>> Steven Hawkings about cosmic inflation: "An increase of billions of
>>>>> billions of percent in a tiny fraction of a second. Of course, that
>>>>> was before the present government."
>>>>> 
>>>>> --
>>>>> Olsr-dev mailing list
>>>>> (spam-protected)
>>>>> https://lists.olsr.org/mailman/listinfo/olsr-dev
>>> 
>>> --
>>> Ferry Huberts
> 
> -- 
> Ferry Huberts





More information about the Olsr-dev mailing list