Thu May 18 13:53:34 1995  Jim Meyering  (meyering@comco.com)

	* Version 1.2.1.

	* src/Makefile.in (dist): Correct so that config directory is
	included.

Tue May  9 00:15:21 1995  Jim Meyering  (meyering@comco.com)

	* Version 1.2.

	* Lots of minor changes related to converting to autoconf-2.

Wed May 25 08:08:47 1994  Jim Meyering  (meyering@comco.com)

	* Makefile.in (check): New target.  To invoke same in src.
	* src/Makefile.in (install): Depend on PROGS instead of install.
	Suggested by Nelson H. F. Beebe (beebe@math.utah.edu).

	* mkinstalldirs: New file.
	* Makefile.in: Use it.

	* install-sh: New file.

Fri Feb  4 12:48:33 1994  Jim Meyering  (meyering@comco.com)

	* Version 1.1.13

	* Makefile.in [install, uninstall, TAGS, clean, mostlyclean,
	distclean, realclean]: Consolidate rules for these targets.
	[dist]: Update to use gzip.

	* src/Makefile.in: Always run tests.
	[dist]: Add commands to put Test and config directories
	in the distribution.

	* system.h: Add dcl of error.

	* configure.in [AC_UNISTD_H]: Remove obsolete macro.
	[AC_HAVE_HEADERS]: Add unistd.h.

	* s2d.1: Update for `--' option introducer.

	* s2d.l, d2s.l: Define yywrap.  It's required by flex-2.4.6.

Tue Feb 23 17:18:15 1993  Jim Meyering  (meyering@comco.com)

        * main.c: Use version_string instead of VERSION macro.
	(usage): Report usage with -- instead of + long option prefixes.

	* Makefile.in: Rebuild Makefile if it's out of date.

Sun Jan 31 12:42:44 1993  Jim Meyering  (meyering@comco.com)

	* main.c: Include <errno.h>.
	[! STDC_HEADERS]: Declare extern errno.
	Remove K&R code from #ifdef __STDC__.
	Use S_ISDIR if it's available.

	* proto.h: Removed.

	* d2s.l, s2d.l [comment start]: Recognize comments that start with
	`^Cc' or anything else that resembles an identifier.  Thanks to
	June - Yub Lee <jylee@kitty.cims.nyu.edu> for pointing out the problem.
	* Update tests to check for this type of comment.

Thu 05 Dec 1991 18:21:57

  * version 1.1.11 offered to Phil Miller <pwm@macaw.jhuapl.edu>

Wed 13 Nov 1991 14:28:32

  * version 1.1.10

Wed 13 Nov 1991 13:41:03

  * flbreak.c: Use fputs instead of printf.

Wed 13 Nov 1991 09:22:03

  * ChangeLog, version.h: Moved description of changes from version.h
  to ChangeLog.

Tue 12 Nov 1991 15:33:21

  * Fixed problems that arose because symbol.c didn't include
  <ctype.h> Thanks to Dan Giesy <dgiesy@scb3.larc.nasa.gov>.

  * Also, Dan reported that Alliant's `make' couldn't handle Makefile.in's
  lack of default rule for building flbreak from flbreak.c, so I added one.

Fri 08 Nov 1991 11:01:08
    Cleaned up INSTALL, README, Makefile.in, added proper version
    numbers (above) instead of things like "1.2-beta".

Thu 10 Oct 1991 07:42:03
    s2d.l: Was appending lower case "d0" to real literals without explicit
    e-notation.  s2d now appends "D0".  Thanks to Juhyeon Yoon
    <yoonj@cs.rpi.edu> for reporting the problem.

Sun Aug 25 22:19:59 1991
    s2d.l: Fixed bug that improperly allowed translation of contents
    of format statements that looked like reals, when the format
    keyword did not begin in column 7.  Reported by Roger Alexander
    <alex@iastate.edu>.

    d2s.l: Didn't treat format statements specially at all and could
    convert some unlikely constructs.  Now format statements are
    completely ignored here, too.

Wed Jul 17 22:44:36 1991
    I wrote the man page for s2d and d2s a couple months ago.
    Added mention of f-ai to man page.  Fixed a few typos in prototypes.

Thu Mar 28 18:14:28 1991
    Added linpack.names (that now includes complex versions of
    linpack subroutines and functions) from Dan Giesy.

Tue Mar 26 09:43:35 1991
    As Dan Giesy suggested, I've added an explanatory section at the
    beginning of README -- there used to be one... I guess I lost the
    original.

Tue Mar 26 09:33:04 1991
    Added test in Makefile/GNUmakefile of idempotence of fljoin/fbreak

Tue Mar 26 09:29:47 1991
    Eric Grosse <ehg@research.att.com> and Daniel P. Giesy
    <dgiesy@scb3.larc.nasa.gov> reported a bug in fljoin
    that caused it not to remove first 6 chars of continued lines.

Mon Jan 21 08:17:12 1991
    In s2d, both keywords "sngl" and "dble" were being dropped with
    the assumption that all variables were of simple scalar type and
    would be converted to type "double precision."  This was an
    incorrect assumption of course -- dble() and sngl() can be applied
    to COMPLEX variables and their omission in translation is an error.
    Now, "dble" is preserved and "sngl" is converted to "dble".
    Thanks again to Kris Hampel for reporting the problem.

Mon Jan 21 08:03:09 1991
    Thanks to Kris Hampel <kris%sst.ph.ic.ac.uk@pucc.PRINCETON.EDU>
    who reported that s2d can modify the contents of some format
    statemtents.  The problem arose with formats like (7d12.3) which
    were converted to (7d12.3d0).  I had code to handle (d12.3) by
    recognizing the preceding "d".  and ignoring the following real
    number.  But prepending the "7" causes the interpretation of 7d12
    as a double and ".3" as a real constant -- which is mistakenly
    converted to ".3d0".  Anyway, now the entire format statement is
    explicitly ignored.  The disadvantage is that in converting code
    with multi-line format statements (with continuation lines) one
    must use flbreak and fljoin.  But it was necessary to do so with
    s2d in any case since the "real" to "double precision" translation
    could easily lead to lines longer than the 72-char limit.

    This has the added effect (disadvantage for some?) that if the
    string "real" appears in a format statement, it will never be
    converted to "double precision."

Sun Dec  2 14:23:36 1990
    flbreak.c: Eric Grosse pointed out that getchar() returns
	an int, not `char'; also need char cast for putchar().
    symbol.c: should assert(!isupper(c)) rather than islower.
	Thanks to walter@cwi.nl (Walter M. Lioen).
    s2d.l: s2d mistakenly converted "2.5" to "2.d05".  I don't
	know how this one got past the tests -- a strong argument
	for more careful maintainance of regression tests.
	Thanks again to Walter for reporting this problem.
    s2d.l: s2d no longer converts "implicit real(a-z)" to
	"implicit dreal(a-z)".  Reported by Walter.
    README: Warn that converters don't properly handle "dimension"
	statements.  Reported by Walter.

Wed Apr 11 15:13:28 1990
    sent version 1.0 to netlib

March, early April
    minor changes to config/config

Sat Feb  3 23:14:10 1990
    added system prototypes so gcc -W -Wall shuts up
    added many-to-one mapping capability (see ./config/README)
    fljoin no longer has hard-coded limit on input line-length

Mon Jan 29 22:29:56 1990
    added GNU GPL

Mon Dec  4 10:25:05 1989
    moved s2d-d2s dependent stuff to init.c;
    now, there's only 1 version of main.o

Around Dec 1, 1989
    version 1.0 pre-release
