CWRU Bash-1.13 patch 3

31 views
Skip to first unread message

Chet Ramey

unread,
Sep 28, 1993, 1:14:05 AM9/28/93
to
This is the third patch to the 09/03/93 distribution of
bash-1.13.cwru. This fixes a number of small bugs, none fatal.

The patch, as always, may be anonymously FTPd from
slc2.ins.cwru.edu (129.22.8.104), in /pub/dist. The file is
bash-1.13.cwru.patch.3; the explanation is README.patch3.

I'm also making available a new version of the `features.texi'
overview. This document has undergone a massive revision -- it
may even be useful now. Please send any problems you find with
it to me, and don't worry about things like `set -o posix' or
$OSTYPE that are not in the current version -- they will be there
in the next full release. The paper is available in texinfo and
postscript formats in ~ftp/pub/bash/features.{texi,ps}.

The manual page has been updated, and is also available in
~ftp/pub/bash in troff and postscript formats as bash.{1,ps}.
The same caveat about features holds.

As always, send me mail with any problems or if you can't FTP the
patch.

Here is README.patch3:
========================================
This patch (bash-1.13.cwru.patch.3) fixes the following problems:

1. Bash now ensures that the readline vi mode comment character is
"#" when put into `Posix.2' mode. [bashline.c]

2. maxpath.h now no longer includes `sysdefs.h', since sysdefs.h is
supposed to be used only by the build process. cpp-Makefile now
passes along -DHAVE_LIMITS_H in CFLAGS to the build process, which
is what maxpath.h wanted. This allows machines.h entries to do
things like #undef HAVE_RESOURCE on machines that have <sys/resource.h>
but lack the support for getrlimit(). [maxpath.h, cpp-Makefile]

3. Asynchronous jobs started when the shell is not started
interactively, in subshells via `( ... )', or when job control is
not active now get /dev/null as standard input. Previously, all
asynchronous jobs started when the shell was not `running
interactively', such as when sourcing startup files, were given
/dev/null. [execute_cmd.c]

4. The M_OS definition for Solaris 2 machines is now `SunOS5'; Solaris
is now defined in SYSDEP_CFLAGS. [machines.h]

5. Dynix machines are now identified as such, rather than as vanilla
USG systems (USG is still defined in SYSDEP_CFLAGS), and undefine
HAVE_ALLOCA -- apparently the system alloca causes core dumps.
[machines.h]

6. Since HP/UX inherits the SunOS libc problems with interrupted YP
function calls freeing memory multiple times, those machines cannot
safely use the GNU malloc. [machines.h]

7. SCO machines have had the default library order changed slightly.
[machines.h]

8. Posix systems without job control, or shells compiled without job
control on Posix systems, need to restore the top-level signal mask
after forking children, and need to set the tty job control signals,
if they exist, to the default values for all children. [nojobs.c]

9. A bug was fixed that caused machines without varargs.h and old-style
varargs support to not null-terminate jobs in cprintf(), which caused
printed commands to overrun and include stray text from previous
commands. [print_cmd.c]

10. If POSIXLY_CORRECT or POSIX_PENDANTIC is in the initial shell
environment, the shell should go into `Posix.2 mode'. [variables.c]

11. When the result of strlen(3) is used directly in a comparison, it
should be cast to an int. [lib/readline/history.c]

12. Recursive keyboard macros are now disallowed. Previously, an
infinite macro execution loop would occur when C-x e was entered
while defining a macro. [lib/readline/readline.c]

13. The tilde include files are now protected against multiple inclusion.
[lib/readline/tilde.h, lib/tilde/tilde.h]

14. The default vi comment character is now `#'. [lib/readline/vi_mode.c]

15. SCO version 4.2 is now recognized and identified as `SCOv4' for the
auto-config process. [support/mksysdefs]

To apply, cd to the bash source directory and type `patch -p0 < patch-file'.

Chet

--
``The good news is that this year two million people will pass through
Washington's new Holocaust Memorial Museum, which will survive the
survivors and be their testimony.''
Chet Ramey, Case Western Reserve University Internet: ch...@po.CWRU.Edu

Reply all
Reply to author
Forward
0 new messages