If you're developing on a pre-ANSI setup and bringing back ANSI C code, things
like this will bite you (and can't necessarily be expected to all be
backwards compat). IIRC on sunos I used to wrapper realloc with a simple
wrapper that checked the first arg for NULL and called malloc in those cases...


James

>
>In article <20021223055606.GA4392@krishna>
>junyoung@netbsd.org writes:
>
>> IMO, it would be to simply close the PR. If we accepted such request,
>> some people soon might request to back out removing __P() in userland
>> programs (and other ANSI changes as well), because they don't compile
>> in pre-ANSI systems. 
>
>This is not related to NetBSD userland.
>I (I'm the author of the PR :) just thought that
>	develop a program with yacc on NetBSD,
>	bring the C file to SunOS 4,
>	and it always cores
>was a little awkward, and might discourage users to use NetBSD as
>a development environment.
>
>(I have fewer number of SunOS 4's around, recently. :)
>
>-- 
>ITOH Yasufumi
>
>
>