149 lines
		
	
	
		
			5.5 KiB
		
	
	
	
		
			Plaintext
		
	
	
	
			
		
		
	
	
			149 lines
		
	
	
		
			5.5 KiB
		
	
	
	
		
			Plaintext
		
	
	
	
Documentation error in STRING-OUTPUT-PORT-OUTPUT.
 | 
						|
Reported 12/19 by Michel.Schinz@studi.epfl.ch
 | 
						|
 | 
						|
 | 
						|
character-gobbling in (record-reader) caused by 'trim / 'peek
 | 
						|
default misunderstanding in delimited readers. Fixed 4/5/96
 | 
						|
Reported by Victor Zandy
 | 
						|
 | 
						|
reap-policy = early can still lose if you loop and fork.
 | 
						|
fork now reaps & retries if it loses and the policy is early reap.
 | 
						|
This is a kludge until I have sigchld handlers.
 | 
						|
Fixed 4/5/96; reported by Michael Becker.
 | 
						|
 | 
						|
CHAR-SET-CONTAINS? now works for non-ASCII chars.
 | 
						|
 | 
						|
SKIP-CHAR-SET now exported and documented. 4/5/96
 | 
						|
 | 
						|
Tod Olson <ta-olson@uchicago.edu> complained about the I/O performance
 | 
						|
    of the delimited reader routines (the supposedly fast ones, with
 | 
						|
    C support). Reported 11/95, improved 4/96.
 | 
						|
    Things generally improved all over by improving the efficiency
 | 
						|
    of optional argument handling (but especially I/O, I think).
 | 
						|
 | 
						|
I changed the meta-arg parser (C, Scheme, and doc) so that newline
 | 
						|
    *always* terminates an arg -- empty or non-empty. KISS.
 | 
						|
 | 
						|
(chdir) was cd'ing to ".", not home. Reported by dmh@tibco.COM (David Hull).
 | 
						|
 | 
						|
autoreap-policy's optional arg wasn't optional -- the dot was missing.
 | 
						|
    Reported by rthappe@mathematik.uni-freiburg.DE (Rolf-Thomas Happe).
 | 
						|
 | 
						|
alist->env didn't work, because it was passing a list, not a vector,
 | 
						|
    to the %INSTALL-ENV procedure. Reported by Mike Sperber 9/19/96.
 | 
						|
 | 
						|
syntactically illegal LET in rw.scm. Reported by jsc 9/23/96.
 | 
						|
 | 
						|
a semantically incorrect but benign error in syscalls.1c.
 | 
						|
Reported by Lewis Girod 12 Dec 95. Fixed by bdc 26 Sept 96.
 | 
						|
 | 
						|
"make clean" didn't get the .o's in scsh/machine
 | 
						|
Reported by Kevin Esler 2 April 96. Fixed by bdc 26 Sept 96.
 | 
						|
 | 
						|
support for Solaris 2 on i386 in configure.in
 | 
						|
Reported by Lutz Euler on 11 Jul 96. Fixed by bdc 26 Sept 96.
 | 
						|
 | 
						|
socket-address->unix-address returns function not address
 | 
						|
Reported by Sean Doran 29 Jul 96. Fixed by bdc 26 Sept 96.
 | 
						|
 | 
						|
Solaris users with BIND installed lose MAXHOSTNAMELEN
 | 
						|
reported by Lutz Euler 23 Aug 96. Fixed by bdc 26 Sept 96.
 | 
						|
 | 
						|
on HP-UX use dld dynamic loader for symbol lookup (faster)
 | 
						|
reported by Bill Sommerfeld 31 Oct 95. Fixed by bdc 27 Sept 96.
 | 
						|
 | 
						|
problem with path marshalling with the unix domain protocol
 | 
						|
Reported by Sean Doran 29 Jul 96. Fixed by bdc 1 Oct 96.
 | 
						|
 | 
						|
exportlist.aix generated correctly on aix 4 now
 | 
						|
Reported by Michael Becker 2 June 96. Fix by Michael Sperber 2 Oct 96.
 | 
						|
 | 
						|
service-info using a port number is broken for little endian machines
 | 
						|
Reported by Jin S. Choi 3 Oct 1996. Fixed by bdc 3 Oct 96.
 | 
						|
 | 
						|
read_fdport_substring() didn't reset the port's peek-char to #F when
 | 
						|
    consuming it. Reported by JSC 10/16/96, with fix.
 | 
						|
 | 
						|
*** 0.4.3 Released by bdc 31 Oct 96
 | 
						|
 | 
						|
problem with solaris 2.5.1 because fileno is not a macro anymore?
 | 
						|
Reported by Roebert E. Brown. Fix by bdc 3 Nov 96
 | 
						|
 | 
						|
distclean should remove emacs ~ backup files not clean
 | 
						|
Reported by Roebert E. Brown. Fix by bdc 1 Nov 96
 | 
						|
 | 
						|
signals-ignored-by-default for nextstep
 | 
						|
Reported by Roebert E. Brown. Fix by bdc 1 Mov 96
 | 
						|
 | 
						|
linux on a pentium pro aka 686 not properly handled by configure
 | 
						|
reported by Douglas S. J. De Couto. Fix by bdc 3 Nov 96
 | 
						|
 | 
						|
problem compiling on sunos 4.1.x
 | 
						|
reported by Franklin Chen. Fix by bdc 3 Nov 96
 | 
						|
 | 
						|
signals-ignored-by-default for nextstep
 | 
						|
Reported by Mike Sperber. Fix by bdc 3 Nov 96
 | 
						|
 | 
						|
*** 0.4.4 Released by bdc 3 Nov 96
 | 
						|
 | 
						|
Makefile.in didn't contain correct scsh.vm target
 | 
						|
Reported by Mike Sperber. Fix by bdc 4 Nov 96
 | 
						|
 | 
						|
Makefile.in needs to rerun ranlib after installing libscshvm.a
 | 
						|
Reported by Robert E. Brown. Fix by bdc 4 Nov 96
 | 
						|
 | 
						|
signals-ignored-by-default for ultrix
 | 
						|
Reported by Jay Nietling. Fix by bdc 4 Nov 96
 | 
						|
 | 
						|
signals-ignored-by-default for irix
 | 
						|
Reported by Richard Kelsey. Fix by bdc 4 Nov 96
 | 
						|
 | 
						|
Starting up programs with the "-e <entry-point>" command-line
 | 
						|
switch and with the (DUMP-SCSH-PROGRAM <file-name> <entry-point>)
 | 
						|
are now equivalent in that both pass a list of command-line arguments
 | 
						|
that includes the program name. The two start-up methods were not
 | 
						|
the same in the previous release. Reported by Dan Hagerty hag@ai.mit.edu
 | 
						|
October 1996.
 | 
						|
 | 
						|
transcendental functions broken (sin, cos, tan, log)
 | 
						|
reported by jsc 8 Nov 96. Fix by ,open floatnum
 | 
						|
 | 
						|
Fixed up initial sizes for allocated filename buffers to be _POSIX_PATH_MAX.
 | 
						|
 | 
						|
Solaris lacking three netconst.scm definitions
 | 
						|
reported by Michel Schinz 15 Nov 96. fix by ???????????
 | 
						|
 | 
						|
Irix 6.x can't compile sighandler1.c for lack of sigbits?
 | 
						|
reported by Stephen Pascoe 18 Nov 96. fix by ??????????
 | 
						|
 | 
						|
host-info:addresses contains bogus info if the argument to 
 | 
						|
host-info is a socket-address.
 | 
						|
reported by Michael Sperber 2 Dec 96. fix by bdc
 | 
						|
 | 
						|
signals1.c missing on most ports
 | 
						|
reported by Mike Gunter 10 Dec 96. fix by ??????????
 | 
						|
 | 
						|
scsh.scm
 | 
						|
    stdio->stdports was using fdes->inport to make the current-output-port
 | 
						|
    and the current-error-port. The manual also had this bug.
 | 
						|
 | 
						|
    The manual also used (exit (run ...)) in section 9.1.7, which should
 | 
						|
    be (exit (status:exit-val (run ...))).
 | 
						|
 | 
						|
    Reported by euler@lavielle.COM (Lutz Euler) 2/24/97.
 | 
						|
    Fixed by Shivers 2/24/97.
 | 
						|
 | 
						|
Alan Bawden has reported a lot of bugs 4/97:
 | 
						|
    Ignoring SIGPIPE is fine for scsh, but screws up sleazy C programs
 | 
						|
    exec'd from scsh. Scsh now runs with SIGPIPE's handler set to a no-op
 | 
						|
    C fun.
 | 
						|
 | 
						|
    user-info C code was returning garbage strings when there was an error,
 | 
						|
    crashing the system.
 | 
						|
 | 
						|
    set-file-group had bugs (uid/gid args were transposed in the syscall)
 | 
						|
 | 
						|
meta.scm was blowing up on illegal numeric escapes like \3Q in meta-arg lines.
 | 
						|
    Was also not recognizing the whole set of C escapes (e.g. \n, \a, \f).
 | 
						|
    Jim Blandy reported the bug, with fixes for the second problem, 4/97.
 |