Regarding your MS-DOS question, do not attempt to open the same file twice for writing, except in the privacy of your own home and unless you have made full backups of your entire disk and you are prepared to reformat the disk.

I can only guess that since MS-DOS was not designed as a multitasking system, it's designers assumed that a file would never be opened twice. However products like DESQview can do some checking along these lines. (I use DESQview extensively as a multitasking environment running over MS-DOS.

My old Wordstar does not have multiple buffers. When I open three or four Wordstar windows to edit multiple files, with fast keystroking I sometimes open up multiple copies of the same file.

I suppose DESQview does some checking for me since I haven't had to reformat the disk. Wordstar creates a couple of temporary files when a document is opened. With multiple copies being edited, the contents in these files may become garbage, but disk allocation (as tested with CHKDSK) does not have problems.

So it is possible to protect against this source of disk corruption — but not with straight MS-DOS. Now I have a question: does anyone know if MS-DOS v4.0 differs in its response to multiple opens?

Q

I have discovered that many short, seemingly simple programs simply do not run correctly until you insert a printf() for debugging purposes. I have noticed that the position of the printf() with respect to the suspect code also affects the outcome. Can you explain this? Thanks.

Mark Petrovic
Stillwater, OK

A

You either have the pointer blues or the compiler bug. The latter is a rather rare item, so I suspect a pointer problem.

Most probably, you are wiping out some code by writing through an uninitialized pointer reference. The addition or deletion of some printf's changes the position of the code causing the damage to occur at a more benign location.

I recommend Robert Ward's book on Debugging C as the best place to start in tracking down these errant pointer problems.

/*  Sample if statement */
if (a   0)
 {
   a = 0;
   b = 1;
 }
else
 {
   a = 1;
   b = 0;
 }

On a related note, I have also given up using tabs (except for a single initial tab) for indentation. I move my code between systems on a fairly regular basis, and it never fails to surprise me just how many ways of handling tabs and spaces there are in various editors and printers. As a result, my indenting style consists of a single tab for the first level of code, and three spaces for each following level (with braces being undented by a single space).

Anyway, I just wanted to provide you with another possible style for future consideration, and to let you know that I think your column is very good.

Brian S. Merson
Marlboro, MA

Thank you for your contribution. Your style has a lot going for it. If pretty-print programs would output it, I think I would prefer it to mine. As far as typing it originally, my fingers would probably refuse to learn it. (KP)

To improve the readability of the if () ... code it's convenient to distinguish between the case when else is present and the case when it is not present. This allows to expect or not expect else (which may appear on the next page) already when reading if () ... statement.

Therefore I use a separate line for an open brace when if () is followed with else. When else is not expected the open brace is on the same line as the controlling statement.

So, the codes look like these:

if (...)  {
   ...
   }
if (...)
   {
   ...
   }
else
   {
   ...
   }

This also has some possibilities. My only objection would be that adding an else to an if, would require going back and changing the line on which the if appears. (KP)

A couple of years ago I had to write a query-utility in C to query the database of a program that calculates salaries and taxes. No information on the MBF-format was available but I was able to reverse engineer it. The format is very similar, but not identical to IEEE.

For example MKD$ creates an eight-byte string. The first byte is a sign bit. Then follows a seven bit exponent followed by seven bytes of mantissa.