Posted by John Gleason on August 09, 2006 at 07:42:06:
In Reply to: Re: Qantel Conversion posted by Marc on August 01, 2006 at 15:07:52:
Hmmmm...
Any code can be disassembled. A decompiler may not be 'cleanly' decompiled, but it can be as easily reverse-engineered as Qantel or Comet code can be decompiled.
There's really no excuse for 'messy' decompiled Qantel or Comet object code (a caveat follows). I've learned through my own experiences decompiling both Qantel (the source code resulting from my decompilations was certainly initially messy!) and Comet code that it's ultimately programmer laziness. There may be no easy way of automating a clean decompile, but having a base set of decompiled source to start with it's relatively easy to manually clean it up. Comet object code is even easier to decompile given how MTB's output provides a very clean and predictable object-code layout and the decompiled code is far cleaner eliminating a lot of the messiness present in Qantel decompiled code.
The aforementioned caveat is the loss of variable names in the decompiled code. The symbol table is not stored in the object code (the symbol table contains information about the variables - their names, lengths and whether they are strings or numerics). The names of constants are also not stored.
Compared to how Qantel used to store object code (perhsps still does), Signature Systems provides quite a few improvements. One example is how Comet pre-relocates code during the compilation process ('relocation' allows the engine running the object code to distinguish between common and local variables).
Ultimately I have to extend my compliments to Jim and his team for a great job they've done with Comet.