src-test2/include/stdio.h, branch release/3.4.0_cvs

This commit was manufactured by cvs2svn to create tag

1999-12-20T00:16:49Z

'RELENG_3_4_0_RELEASE'. This commit was manufactured to restore the state of the 3.4-RELEASE image. Releases prior to 5.3-RELEASE are omitting the secure/ and crypto/ subdirs.

$Id$ -> $FreeBSD$

1999-08-29T14:49:43Z

MFC: fseeko() and ftello().

1999-05-05T22:14:47Z

Make this protototype of mmap return void *, like two other copies.

1998-12-14T19:23:16Z

Added macros printflike() and scanflike() to .

1998-07-13T06:45:23Z

Use them to `make gcc -Wformat' check formats for all printf-like and scanf-like functions in /usr/src except for the err()/warn() family. err() isn't quite printf-like since its format arg can legitimately be NULL. syslog() isn't quite printf-like, but gcc already accepts %m, even for plain printf() when it shouldn't.

New flag for internal stdio use to enable dynamic string allocation for

1998-07-08T00:52:40Z

asprintf() implementation. Obtained from: Todd Miller via OpenBSD

Complete the switch to using gcc-2.4 features to declare fpos_t ,

1998-06-14T16:04:20Z

as threatened in the 4-5 year old comment. Most of the change is in ; this commit just removes cruft for the strict ANSI case. 64-bit types couldn't actually be defined using __attribute__(()) in gcc-2.4, since attribute names in the implementation namespace only became available in gcc-2.7. I've probably just broken gcc-2.6 support by using the gcc-2.7 names unconditionally.

I have a long list of issues to address here, mostly related to

1998-05-05T21:59:19Z

namespace and revision levels of ANSI and POSIX. This change only removes the leading underscrore from the FILE locking functions that POSIX defines.

Format changes in an attempt to address some of Bruce's comments

1998-04-12T20:29:24Z

about spaces and tabs. The externs in the static inline functions remain 'cause (a) they're required; and (b) I can't find any gcc -W* cases where they generate warnings.

Add *_unlocked() macros according to POSIX. In the _THREAD_SAFE case,

1998-04-11T07:33:46Z

test for __isthreaded before calling the lock/unlock functions to try to save some performance. The _THREAD_SAFE case should become the default, but since it tests for a global variable in libc, people won't be able to build -current on pre-3.0 systems unless the default leaves it out. Such is life.