Re: char_special_data

From: Chris Field (chrisf@ptd.net)
Date: 09/27/99


At 12:14 AM 9/28/99 CEST, you wrote:
>This seems to be similar to the problem i am having too,
>when mud crashes when using olc... i can add 2 chars, but
>when addin the third the mud crashes.. (using olc)
>
>
>shrug
>
Hmm whenever I've had crashes from adding new structures/variables it is
because I'm too tired and I get sloppy.  First off, if you are using binary
pfiles and a save structure...wipe the pfiles and make sure anything added
to the save structure is fixed length.  Next, make sure you are
initializing any char * fields to null in init_char(or any pointer fields
that will be empty).  If you want to copy something to the pointer, check
if it is null first...then initialize the memory.  When you free a pointer
after the free(bleh) set it to NULL because free doesn't automatically do
that.  Make sure when you load/save chars you throw some sort of dummy
character to signify the field is empty and can be set to null, unless you
have an ascii pfile format that initializes all the structures and pointers
and variables to NULL or 0 or whatever a good default value is, and then
scans the pfile and fills in the fields that are entered into the pfile(the
really spiffy way of having things...)

pretty much ch->desc is null, but it can be null for many reasons.  My
guess is that instead of descc you put desc in one of your frees.  Also if
you have overwritten the memory taken up by d->olc, by writing to a pointer
that is pointing to unallocated memory, or already allocated memory then
doing an if on part of that memory that is written weird and totally
invalidated can do weird stuff that right now I am too tired to describe :P
 An error in a seemingly distant part of the mud with regard to writing on
memory that wasn't intended to be written on with a pointer can affect
variables that are files away.

Therefore, I suggest a good logging system for changes(which hopefully you
have).  Then you can say...okay my mud was working before I added this,
that, these and those, and yet after adding thus and that I remember it
working, but these never quite worked 100%.  Therfore you recall what the
these project was and go over everything by rereading(maybe even printing
and deskchecking *cough*).

To avoid problems like this, I am going to give the greatest piece of
advice I have ever heard as regards programming :P  When you first write a
subprocedure to do something, you test it thoroughly and make sure it does
what it is supposed to do and only what it is supposed to do.  In fact,
read over it a few times to be totally sure.  Then once you know that sub
procedure is complete you can make another sub procedure that includes the
previous one with confidence that any errors in the new sub procedure will
be in the new sub procedure.  Then, just keep testing up the changs and
including subprocedures into other subprocedures until things are verified
and then error checking will be swift.  Of course, this procedure is not
always feasible when modifying an already written program, but my idea for
implementing the procedure would be like this.....

1.  Add the new variables and add a procedure to set it to a value.  Also,
set the values to null when they are supposed to be null.  Then, just try
to see if you can change the values with an online command.
2.  If this works then add in things to save the values and use an online
command to change the values and then try saving and quitting and seeing
that they change
3.  Okay everything has gone perfectly.  Just add in the new menu
screen(thoroughly testing it) and have it call the procedures to set the
values for the new variables.
4.  Check everything is working, remove any procedures that unnecessarily
display the values for the variable and you are done(almost)
5.  Add places in the code where you want to use the value of the
variable(using a similar procedure to test them)

following this procedure will probably not steer you wrong.  Admittedly I
do not follow this procedure because I am lazy :P  But anyway, following
this procedure and constantly testing things could lead to the rock solid
mud of legend.  You could be the envy of the entire circlemud community!!!

                                                Oh well enough of this tiresome arbitrary gossip and time to get back
to some even more tiresome arbitrary constants!!!!

From,
Cervo Ovrec The Beaten..........


     +------------------------------------------------------------+
     | Ensure that you have read the CircleMUD Mailing List FAQ:  |
     |  http://qsilver.queensu.ca/~fletchra/Circle/list-faq.html  |
     +------------------------------------------------------------+



This archive was generated by hypermail 2b30 : 12/15/00 PST