[ntp:hackers] Re: header collisions

David L. Mills mills at udel.edu
Tue Sep 20 13:01:40 UTC 2005


Harlan,

I have no idea what is going on here. How does the Linux kernel undefine 
anything? Why is it involved with compiling NTP? There might be 
conflicting header files, but how can Linux undefine anything? Consider 
the gnu debacle where gnu simply replaced the system header files.

As you know, I am very, very violent in Linux criticism. I would really 
rather tell Linux to keep to NTPv3 and not bother us. Yes, I know that 
is impractical, but I really bristle when some Linux offense affects 
defines that have been in NTP for a decade or more. What is the prospect 
of this happening again? I think it better to fix individual cases, like 
rename CONFIG_PHONE to CONFIG_PHONE2. Either that or make sure Linux 
header files go first and undefine/redefine individual cases.

Dave

Harlan Stenn wrote:

>Dave,
>
>The linux kernel overrides and #undef's CONFIG_PHONE and one other
>CONFIG_ variable that we use, and this prevents us from compiling.
>
>I don't think we're gonna get them to change...
>
>H
>--
>  
>
>>Harlan,
>>
>>Please don't do this; I have a hard enough time as it is remembering 
>>which prefix names do what and I do change the suffix names from time to 
>>time. But, I am really not interested in a wholesale change. What 
>>problems do the current naming conventions cause?
>>
>>Dave
>>
>>Harlan Stenn wrote:
>>
>>    
>>
>>>Dave,
>>>
>>>We're getting more reports of problems with the names of the CONFIG_*
>>>variables in ntp_config.h .
>>>
>>>I was gonna rename all of them from CONFIG_* to NTP_CF_* (which happens
>>>to keep the length of the beasts the same).
>>>
>>>Any objection?
>>>
>>>H
>>>
>>>PS: see https://ntp.isc.org/bugs/show_bug.cgi?id=467 for the report.
>>>      
>>>
>>    
>>



More information about the hackers mailing list