[haiku-development] Re: Duplicate headers?
- From: James Leone <linuxcpa@xxxxxxxxx>
- To: haiku-development@xxxxxxxxxxxxx
- Date: Fri, 29 Aug 2014 23:25:32 -0700
"Such trivialities are not specific to Haiku, there is plenty of
documentation elsewhere on #include and command line flags to GCC"
If the BSD headers are ignored otherwise, I guess that means when FreeBSD
wifi drivers are configured, an -I flag will be necessary.
OK, makes sense I suppose.
On Fri, Aug 29, 2014 at 11:18 PM, James Leone <linuxcpa@xxxxxxxxx> wrote:
> "Based on the order of include directory listing on the command line and
> whether there was a bsd/ in the include directive."
>
> So because the output from gcc -v has:
> .../posix
> .../bsd
>
> the 9 BSD headers will always be ignored?
>
>
> On Fri, Aug 29, 2014 at 11:11 PM, James Leone <linuxcpa@xxxxxxxxx> wrote:
>
>> gcc -v shows the following relevant directories as being root to include
>> statements:
>>
>> /boot/system/develop/headers/posix
>> /boot/system/develop/headers/bsd
>> (Among several others)
>>
>> So when source code has "#include <stdio.h>" in it, when the source code
>> compiles the compiler will look in both directories for stdio.h.
>> Similarly, if the #include statement has #include <sys/wait.h> the
>> compiler will look in both /boot/system/develop/headers/posix/sys and
>> /boot/system/develop/headers/bsd/sys for wait.h.
>>
>> We have the following files in both the
>> /boot/system/develop/headers/posix and /boot/system/develop/headers/bsd
>> directories:
>>
>> errno.h
>> time.h
>> stdio.h
>> stlib.h
>> string.h
>>
>> We have the following files in both the
>> /boot/system/develop/headers/posix/sys and
>> /boot/system/develop/headers/bsd/sys directories:
>> wait.h
>> cdefs.h
>> param.h
>>
>> My question was how does the compiler know what to pick, because, from
>> what I understand, they are duplicated.
>> I wasn't asking about the purpose of the BSD headers.
>> If this is an issue - certainly I should mention it - and perhaps get
>> the question resolved for others that are interested to know how the
>> compiler chooses the correct header.
>>
>> The zlib header in our system is missing lines 1350-1352 when compared to
>> the same header present in Linux, with the same authors:
>> Jean-loup Gailly and Mark Adler (zlib.h)
>>
>> The missing(?) code is as follows:
>>
>> 1349 ZEXTERN int ZEXPORTVA gzprintf Z_ARG((gzFile file, const char
>> *format, ...))
>> 1350 #ifdef __GNUC__
>> 1351 __attribute__((__format__(__printf__,2,3)))
>> 1352 #endif
>> 1353 ;
>>
>> This may have been removed for a good reason. I don't know. But if its a
>> bug I should mention it.
>> The only reason I looked at it was because I had a compilation error
>> arise for the program Archangel.
>> I wanted to see if there was a difference given that it is not a header
>> written by Haiku.
>>
>>
>>
>>
>> On Fri, Aug 29, 2014 at 11:38 AM, Augustin Cavalier <
>> waddlesplash@xxxxxxxxx> wrote:
>>
>>> On Aug 29, 2014 11:48 AM, "James Leone" <linuxcpa@xxxxxxxxx> wrote:
>>> >
>>> > I noticed that some code that includes headers i.e. <*.h> wants to
>>> read the /boot/system/develop and .../develop/posix and .../develop/bsd
>>> directorories. Similarly, <sys/*.h> would read from both of these:
>>> ....posix/sys and .....bsd/sys.
>>> >
>>> > We have 9 duplicates amongst theze directories.
>>> >
>>> > How does the compiler know which one to pick?
>>>
>>> Based on the order of include directory listing on the command line and
>>> whether there was a bsd/ in the include directive. And these are nor
>>> duplicates, they are for the freebsd WiFi compat layer.
>>>
>>> Such trivialities are not specific to Haiku, there is plenty of
>>> documentation elsewhere on #include and command line flags to GCC.
>>>
>>> > P.s. our zlib.h is misding 3 lines of code compared to Linux' same.
>>> Same author and pub date.
>>>
>>> Which are?
>>> -Augustin
>>>
>>
>>
>
Other related posts:
