01-BUGS.TXT  -                www.merlyn.demon.co.uk     >=2007-02-13

1996-12-20 : This file is for reports on any bugs which may be found
in programs on this site.  It assumes that major comment in source
files has been read and heeded.


1998-02-17 : As I don't have a fast PC, those programs with "uses Crt"
invoked will fail RTE200 on fast machines >= 200MHz.  Some or all
since fixed.

1998-07-30 : LONGMULT - I discovered that extended Carry is imperfectly
handled; it does not affect the result, but LONGMULT is now withdrawn
as LONGCALC is much more versatile and is still programmed for F9.

1998-10-09 : LONGCALC - From 1998-10-04 or thereabouts up until today,
a duplicated line in Fsbs caused error if the command "sbs" was used.

1998-10-09 : I have a fast PC, but still use the older one for Pascal;
the earlier comment still applies.
(ff. : unless TPPATCH applied; ff. which I normally now do)

1999-06-06 : LONGCALC - Within the last few weeks, it has been able to
divide with negative numbers.  Until spotted yesterday, division of a
numerator by a denominator of the opposite sign and one or more digits
longer would give a range-check error (215).  Fixed.

1999-08-28 : DATEPROX - Valid_Date corrected for Year 0.

1999-08-31 : LONGCALC - May err on Easter; if date is one of Apr 18/25,
gets the other.  Fixed 1999-09-01.

2000-12-14 : LAGRANGE - corrected for "CofG" & R<>1.

2001-09-24 : HYtoMJD is suspect in DATEPROX + PASSOVER (Recent code).
Code withdrawn from DATEPROX, PASSOVER withdrawn.  P.S. correct code
now in HEBCLNDR.PAS.

2001-10-25 : COLS - limit on ! parameter is 48, not the intended 127,
Thinking about best solution. New limit to be 2^31-1.  OK 2001-10-26.

2001-11-30 : Don't use MASSEXEC in Win9x (and up?); an OS bug
drops occasional characters. OK in DOS 6.20. Use COLS > $$$.BAT instead.

2003-01-16 : SUBS : limit of 32K characters removed.

2003-08-31 : TZ-CHECK.  Bias, needed away from UK, was not properly
included.

2005-04-28 : LAGRANGE - Error in L2 for "CofG" fixed.

2005-07-22 : noticed and fixed that LONGCALC 'bas' conversion ignored sign.

2007-02-13 : MJD_DATE gave error described in WEEK-CAL.TXT; some #53-#05
should be #53-#04.  Fixed.

-- 
  JRS.