BUGS file

BUGS file

[Zoltan Hidvegi]

The BUGS file says that there are problems on ULTRIX with gcc.  The problem is
really in nawk.  As I remember, the latest BUGS file did not contain this.

The patch below also removes the multi-io bug/feature description since Zefram
fixed it by addin NO_MULTI_IOS option (which is not yet part of the baseline).

The MACHINES file contains the note about SunOS 4.1.3 (which is referenced
from the BUGS file).

Cheers,

   Zoltan

*** Etc/BUGS	1995/11/07 18:15:16	1.1
--- Etc/BUGS	1995/11/07 18:15:30
***************
*** 1,15 ****
  -*- text -*-
  ------------------------------------------------------------------------
- % ( echo foo ; echo bar 1>& 2 ) 2>&1 1>/dev/null | cat
- bar
- foo
- % ksh
- $ ( echo foo ; echo bar 1>& 2 ) 2>&1 1>/dev/null | cat
- bar
- 
- This is not exactly a bug, but an effect of the multiple IO
- redirection.  It is a (k)sh incompatibility, though.
- ------------------------------------------------------------------------
  Completion has a habit of doing the wrong thing after a
  backslash/newline.
  ------------------------------------------------------------------------
--- 1,5 ----
***************
*** 60,70 ****
  and ksh and may be removed in the future. A good fix would be to keep
  such patterns unchanged if they do not match regardless of the state of
  the nonomatch and nullglob options.
- ------------------------------------------------------------------------
- On ULTRIX 4.2, if NOTIFY is unset, zsh gets SEGV each time a background
- job terminates if compiled with gcc-2.6.3 or c89. It does not happen
- with cc (non-ANSI compiler, ansi2knr is used). Terminal suspension
- does not seem work regardless of the compiler.
  ------------------------------------------------------------------------
  On some systems nawk is broken and produces incorrect signames.h file.
  This make the signals code unusable. This often happens on Ultrix, HP-UX,
--- 50,55 ----
*** Etc/MACHINES	1995/09/20 10:53:32	1.1
--- Etc/MACHINES	1995/11/07 18:22:41
***************
*** 39,46 ****
  
  Sony:	NWS with NEWS-OS 3.3
  
! SUN:	SunOS 4.1.1, 4.1.3
  		Builds ok with native cc and gcc 2.5.8.
  	Solaris 2.1, 2.2, 2.3, 2.4
  		Builds ok with both gcc 2.5.8, and SPARCompiler 3.01.
  		When building zsh on Solaris, you should unset your
--- 39,56 ----
  
  Sony:	NWS with NEWS-OS 3.3
  
! SUN:	SunOS 4.1.1, 4.1.2, 4.1.3
  		Builds ok with native cc and gcc 2.5.8.
+ 		Under 4.1.3 if yellow pages is used, username completion
+ 		may cause segmentation violaton.  This is a bug in the
+ 		shared library not in zsh.  Some libc.so.1.9.x has this
+ 		bug (it fails in yp_all).  Statically linked binaries will
+ 		work if linked with libc.so.1.8.1 (which means that if
+ 		you can get a statically linked binary compiled under 4.1.2
+ 		that it will probably work).
+ 		An altermative but untested solution may be to undefine
+ 		HAVE_NIS in config.h.  This may work, but the first username
+ 		completion will be _very_ solw (as slow as in tcsh).
  	Solaris 2.1, 2.2, 2.3, 2.4
  		Builds ok with both gcc 2.5.8, and SPARCompiler 3.01.
  		When building zsh on Solaris, you should unset your

Re: BUGS file

[Dave Sainty]

Zoltan Hidvegi writes:

> ------------------------------------------------------------------------
> On some systems nawk is broken and produces incorrect signames.h file.
> This make the signals code unusable. This often happens on Ultrix, HP-UX,
> IRIX (?). Install gawk if you experience such problems.
> ------------------------------------------------------------------------

This seems to pop up regularly enough to make it worth kludging... A
detect of the problem should be easy enough to at least point out the
problem exists on machine X. Perhaps even a script that _works_ on the
broken nawk's could be produced....

Unfortunately all the nawks at hand seem to produce identical results
to gawk. Shame, I was looking for a reason to put off the work I
should be doing... :-)

Dave.

BUGS file

[Zoltan Hidvegi]

Here is the latest version from beta9-hzoli11.  The cross ref. to Etc/MACHINES
is missing from test9.  It should mention, that yp_all is broken on some SunOS
4.1.3 systems (e.g. the one in libc.so.1.9.1 is broken).

Zoltan

-*- text -*-
------------------------------------------------------------------------
Completion has a habit of doing the wrong thing after a
backslash/newline.
------------------------------------------------------------------------
If you suspend "man", zle seems to get into cooked mode.  It works ok
for plain "less".
It is not specific neither to man nor to zsh.
E.g. call the fallowing program foo:
#include <sys/wait.h>
#include <unistd.h>

void main(int argc, char *argv[])
{
    int status;

    if (!fork())	/* child */
	execvp(argv[1], argv + 1);
    else		/* parent */
	wait(&status);
}
Then if you suspend
% foo less something
from zsh/bash, zle/readline gets into cooked mode.
------------------------------------------------------------------------
% test=test
% echo "${test##`echo '*'`}"
test
sh gives empty string here. Is it a bug or a feature?
------------------------------------------------------------------------
% zsh -c 'cat a_long_file | less ; :'
can be interrupted with ^C. The prompt comes back and less is orphaed.
If you go to the end of the file with less and cat terminates, ^C
will not terminate less. The `; :' after less forces zsh to fork before
executing less.
------------------------------------------------------------------------
On Suns in xterm if a line wraps to the next line and the cursor is in
the last column on the sreen and you type a character the charactr after
the cursor is replaced on the screen with the character the cursor is on.
------------------------------------------------------------------------
An other problem which may be related to the previous one. In Sun xterms
the output of the following is overwritten by the next prompt if
no_prompt_cr is not set, and if it is set the first character of the
prompt appears in the last column overwriting the `9' character there:
% echo 01234567890123456789012345678901234567890123456789012345678901234567890123456789 | more
------------------------------------------------------------------------
The pattern %?* maches names beginning with %? instead of names with at
least two characters beginning with %. This is a hack to allow %?foo job
substitution without quoting. This behaviour is incompatible with sh
and ksh and may be removed in the future. A good fix would be to keep
such patterns unchanged if they do not match regardless of the state of
the nonomatch and nullglob options.
------------------------------------------------------------------------
On some systems nawk is broken and produces incorrect signames.h file.
This make the signals code unusable. This often happens on Ultrix, HP-UX,
IRIX (?). Install gawk if you experience such problems.
------------------------------------------------------------------------
On some systems (e.g. on Solaris or on AIX)
% zsh -c ':' &
gives
[1]  + suspended (tty output)  zsh -c ':'
This means that zsh scripts cannot be started in the background.  They
can still be backgrounded if started without &, stopped by ^Z and
backgrounded by bg.  The tty output signal is probably comes from the
initialisation of the TERM variable.
% zsh -c ':' < /dev/null >& /dev/null &
works fine.
------------------------------------------------------------------------
Username completion may cause SEGV on SunOS 4.1.3 and NIS.
This is not a zsh bug.  See Etc/MACHINES for details.
------------------------------------------------------------------------