Hello Ingo, On 02/24/2014 01:37 AM, Ingo Schwarze wrote: > Hi, > > i just noticed that the stpcpy(3) manual contains a speculation > that appears to be untrue on closer investigation: That function > did not originate in MS DOS, but in Lattice C on AmigaDOS. > > Here is a patch against the git master HEAD to fix that, and add > some more historical information. To provide some background and > allow you to more easily verify the correctness of the patch, i'm > appending my mail to <misc@xxxxxxxxxxx>, where i'm giving some more > details about the history and pointing to some primary sources. > That mail also contains the (similar, but shorter) patch i just > committed to the OpenBSD manual page. > > Yours, > Ingo > > diff --git a/man3/stpcpy.3 b/man3/stpcpy.3 > index 4718a96..1952df1 100644 > --- a/man3/stpcpy.3 > +++ b/man3/stpcpy.3 > @@ -76,9 +76,10 @@ function is thread-safe. > .SH CONFORMING TO > This function was added to POSIX.1-2008. > Before that, it was not part of > -the C or POSIX.1 standards, nor customary on UNIX systems, but was not a > -GNU invention either. > -Perhaps it came from MS-DOS. > +the C or POSIX.1 standards, nor customary on UNIX systems. > +It first appeared in the Lattice C AmigaDOS compiler (1986 or earlier), > +then in the GNU fileutils and GNU textutils in 1989, > +and in the GNU C library until 1992. > It is also present on the BSDs. > .SH BUGS > This function may overrun the buffer > diff --git a/man3/stpncpy.3 b/man3/stpncpy.3 > index 60042e6..5bc084d 100644 > --- a/man3/stpncpy.3 > +++ b/man3/stpncpy.3 > @@ -96,6 +96,7 @@ function is thread-safe. > .SH CONFORMING TO > This function was added to POSIX.1-2008. > Before that, it was a GNU extension. > +It first appeared in version 1.07 of the GNU C library in 1993. > .SH SEE ALSO > .BR strncpy (3), > .BR wcpncpy (3) > > P.S. > I'd like to apologize that our linker is using strong language > to express its opinion regarding these functions; back in 2012, > the developer implementing them wasn't exactly amused that they > finally made it into POSIX... ;-) Thanks for this note. Now, leaving aside the fingerpointing ;-).... the patch seems more or less okay to me, but given that The ImagaDOS compiler is the merely the oldest instance you could find, I'd prefer a slightly more open wording. How would this be: [[ --- a/man3/stpcpy.3 +++ b/man3/stpcpy.3 @@ -76,9 +76,11 @@ function is thread-safe. .SH CONFORMING TO This function was added to POSIX.1-2008. Before that, it was not part of -the C or POSIX.1 standards, nor customary on UNIX systems, but was not a -GNU invention either. -Perhaps it came from MS-DOS. +the C or POSIX.1 standards, nor customary on UNIX systems. +It first appeared at least as early as 1986, +in the Lattice C AmigaDOS compiler, +then in the GNU fileutils and GNU textutils in 1989, +and in the GNU C library until 1992. It is also present on the BSDs. .SH BUGS This function may overrun the buffer ]] ? Cheers, Michael PS It does seem sad that it ended up in POSIX. I wonder how that came about. > ----- Forwarded message from Ingo Schwarze <schwarze@xxxxxxx> ----- > > From: Ingo Schwarze <schwarze@xxxxxxx> > Date: Sun, 23 Feb 2014 23:29:19 +0100 > Cc: misc@xxxxxxxxxxx > Subject: Re: while trying to compile gettext 0.18.3.2 I see questionable > messages > > Hi Lorenzo, > > Lorenzo Beretta wrote on Sun, Feb 23, 2014 at 05:29:54PM +0100: >> dc@xxxxxxxxxxx wrote on Sun, Feb 23, 2014 at 08:54:34AM -0500: > >>> ../gnulib-lib/.libs/libgettextlib.so: warning: >>> stpcpy() is dangerous GNU crap; don't use it > >> Yet stpcpy(3) on linux says: >> CONFORMING TO >> This function was added to POSIX.1-2008. Before that, it was >> not part of the C or POSIX.1 standards, nor customary on UNIX >> systems, but was not a GNU invention either. Perhaps it came >> from MS-DOS. It is also present on the BSDs. > > I checked the following without finding any references to stpcpy: > - AT&T UNIX v3 to v7 including PWB and 32v > - System III AT&T Unix > - all versions of CSRG BSD from 1BSD to 4.4BSD-Lite2 including SCCS > > The earliest occurrence of stpcpy() i was able to find was in > the Lattice C AmigaDOS Compiler Version 3 Programmers Reference > Guide (1986-09-12)(Lattice Inc.), which explicitly classifies the > function as "TYPE: LATTICE", see > https://archive.org/details/Lattice_C_AmigaDOS_Compiler_Version_3_Programmers_Reference_Guide_1986-09-12_Lattice_Inc. > > So the claim by Terry Lambert that it originated in Borland Turbo C > is definitely untrue. The first release of Borland Turbo C > happened in May 1987. I can confirm it was in Borland Turbo C 2.0 > in 1989, though. > > The 386BSD 0.1 release contains GNU textutils-1.3 and GNU fileutils-3.2 > which both contain a file lib/stpcpy.c with a 1989 FSF Copyright. > This is confirmed by looking at the initial commits of the git > history of the GNU coreutils package. > > The function stpcpy() was contained in the initial git import of > glibc on Feb 18, 1995. The file string.texi says "@comment Unknown > origin" at this point in time. The file sysdeps/generic/stpcpy.c > says "Copyright (C) 1992 Free Software Foundation, Inc." at this > point in time. The ChangeLog reports a bugfix to the function on > Jan 7, 1992 by Roland McGrath. > > In the BSDs, here is when it appeared: > - FreeBSD: Oct 3, 2002 by obrien@, written himself > - DragonFly: Apr 7, 2009 ported by pavalos@ from FreeBSD > - NetBSD: May 1, 2009 ported by perry@ from FreeBSD > - OpenBSD: Jan 17, 2012 by kettenis@, reluctantly written himself > > The function stpncpy first appears in glibc with a 1993 FSF Copyright; > according to the ChangeLog, it was introduced on Oct 29, 1993 by > Roland McGrath, and according to the NEWS file, it was first released > with Version 1.07. > > So, to summarize, the Linux manual is rather misleading. Even > though stpcpy() indeed wasn't a GNU invention, it was first introduced > into the UNIX world by extensive use in the GNU coreutils (then > called fileutils and textutils) in 1989, only very few years after > its original appearance. The glibc was the first UNIX-like C library > to include it only three years later, nearly a decade before FreeBSD, > about 15 years before POSIX, and more than two decades before OpenBSD > reluctantly followed, forced by POSIX. That said, > > stpcpy() is dangerous DOS crap; don't use it > > would be slightly more accurate, but given who pushed it during the > early years, > > stpcpy() is dangerous GNU crap; don't use it > > isn't that far off the mark, either. > > Anyway, we should update our manual, see below. > > Yours, > Ingo > > > Index: stpcpy.3 > =================================================================== > RCS file: /cvs/src/lib/libc/string/stpcpy.3,v > retrieving revision 1.5 > diff -u -p -r1.5 stpcpy.3 > --- stpcpy.3 25 Sep 2013 21:50:18 -0000 1.5 > +++ stpcpy.3 23 Feb 2014 22:06:06 -0000 > @@ -174,9 +174,11 @@ and > functions conform to > .St -p1003.1-2008 . > .Sh HISTORY > -The > +The function > .Fn stpcpy > -and > +first appeared in the Lattice C AmigaDOS compiler (1986 or earlier). > +The function > .Fn stpncpy > -functions first appeared in > +first appeared in the GNU C library version 1.07 (1993). > +Both functions have been available since > .Ox 5.1 . > > ----- End forwarded message ----- > -- Michael Kerrisk Linux man-pages maintainer; http://www.kernel.org/doc/man-pages/ Linux/UNIX System Programming Training: http://man7.org/training/ -- To unsubscribe from this list: send the line "unsubscribe linux-man" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
