Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!usc!elroy.jpl.nasa.gov!sdd.hp.com!wuarchive!uunet!stanford.edu!neon.Stanford.EDU!egret.Stanford.EDU!espie
From: espie@egret.Stanford.EDU (Marc Espie)
Newsgroups: comp.sys.amiga.audio
Subject: Re: Experiment IV
Message-ID: <1991May23.180254.24419@neon.Stanford.EDU>
Date: 23 May 91 18:02:54 GMT
References: <6547@vela.acs.oakland.edu>
Sender: news@neon.Stanford.EDU (USENET News System)
Organization: LIENS, ENS, 45 rue d'Ulm, Paris (France)
Lines: 35

In article <6547@vela.acs.oakland.edu> lmbailey@vela.acs.oakland.edu (Laurana Bailey) writes:
>
>I have a question on Experiment IV? How stable is it?
>
Pretty much stable, bare improvements to come :-).
>The reason I ask is because I listen to alot of MOD files and quite
>often will move them to casette tape for enjoyment in the car. EVERY
>tracker player I have used will CRASH after loading between 10-15
>different modules. I have used Intuitracker (Latest version which
>fixed many a bug), Module Master, UltraTracker, and more recently
>XTPlay. ALL of them would crash after loading 10-15 modules almost as
>though they weren't releasing memory properly. XTPlay works the best
>out of the lot and has a CLEAR button to clear memory, and if used,
>will load between 20-25 modules before crashing. 
It has proved able to load the ~150 modules on ab20 in one round without
crashing. 
>
>These crashes occur even though the memory appears to be released
>fine. Is this a flaw in the MOD format or in the programs themselves?
>
Well, considering that the mod format is a simple binary dump, you have
to be VERY conservative when loading these beasts. The mod format doesn't
include any error-checking, you have to add yours.
Anyway, it's also a pretty non-sensical format. Guess that comes from 
adding kludge after kludge to a small assembly program without ever
designing anything 1/2 :-).

About memory recovery: I've used a personal technique to keep track of
any allocations in a sensible way. Nothing revolutionnary, but it DOES
appear to be much simpler to debug than other techniques.

I guess that checking the program with the Enforcer/memoration/mungwall
did help too !
----
	Marc Espie (espie@flamingo.stanford.edu)