Re: PR#47922

[christos%astron.com@localhost (Christos Zoulas)]

In article <201306171027.r5HARJw7029848%ms-omx11.plus.so-net.ne.jp@localhost>,
Takahiro HAYASHI  <t-hash%abox3.so-net.ne.jp@localhost> wrote:
>Hello list,
>
>On Mon, 17 Jun 2013 15:17:29 +0900
>Masanobu SAITOH <msaitoh%execsw.org@localhost> wrote:
>
>>  Hi.
>> 
>>  I got the same problem as PR#47922:
>> 
>>       http://gnats.netbsd.org/cgi-bin/query-pr-single.pl?number=47922
>
>me too.
>I am trying to build NetBSD/amd64 6.1 on NetBSD/amd64 current(6.99.21).
>and I see similar problem about "_ctype_tab" in
>src/external/mit/xorg/bin/bdftopcf.

The _ctype_tab issue is because of crap left over in the tools nbinclude;
remove all ctype* headers from there and restart you build (perhaps we
should mention it in UPDATING?!?)

As for the __explicit_bzero and __consttime_bcmp(), I still think that
they should be __explicit_memset() and __consttime_memcmp() (the clue
would be that bzero() and bcmp() belong in <strings.h> not <string.h>).

But writing a __consttime_memcmp() is harder than writing a
__consttime_bcmp(), so we punted :-(

I still think that there is time to fix them since they:

1. Don't belong in any formal release and can be removed from the headers
   and left in libc for binary compatibility where the new symbols are
   provided.
2. The API's they replace are documented as legacy and obsolete.
3. They are currently undocumented.

christos