Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!linus!philabs!cmcl2!floyd!vax135!ariel!houti!hogpc!houxm!mhuxa!mhuxi!cbosgd!ihnp4!ixn5c!inuxc!pur-ee!CSvax:Pucc-H:Physics:piner
From: CSvax:Pucc-H:Physics:piner@pur-ee.UUCP
Newsgroups: net.micro
Subject: Yet another software warning.
Message-ID: <798@pur-phy.UUCP>
Date: Tue, 28-Jun-83 04:00:16 EDT
Article-I.D.: pur-phy.798
Posted: Tue Jun 28 04:00:16 1983
Date-Received: Fri, 1-Jul-83 06:33:01 EDT
Lines: 23


  Wow, this bug is really neat. Microsoft strikes again.
Edit-80 assumes a logical record length of 128 bytes.
No matter what the real logical record length is. Under
LDOS (also NEWDOS and other TRS-80 operating systems I
assume.) The logical record length can be anything from
1 to 256 bytes. This is stored in the directory when the
file is created. When you open a file, LDOS will use the
logical record length that you give it. Edit-80 doesn't 
do this correctly, so if the file was created with 256
byte logical record length, Edit-80 opens it with that
but thinks it is working with records of 128 bytes. The
result is that half of each read is lost. As long as
you are only working with files that Edit-80 created in
the first place, you are ok, but if you are working with
ascii files copied from somewhere else, you can have trouble.
LDOS does have a way to change the logical record length
by copying a file, so you can recover the data in a form
Edit-80 will read, but beware of this bug. That is what 
happens when you hack a CP/M program to run under another
OS. Caveat emptor, y'all.
				Rich Piner
				Purdue Physics Dept.