I have been doing dirty patching for the longest time now (at least patched over 3000+ times or so, I patch about 50~ times a day, mostly small iterative changes each time) and have yet to experience any kind of problem / glitches. I have tested with various patched isos that include assembly hacks as well. By patching I mean using the latest FFTPatcher, FFTactext and changing sprites / palettes with ShiShi. I have yet to experience a single glitch. With the exception when using FFTactext; I always load the patch from the file (rather than importing over from the ISO). They run absolutely flawlessly with PCSX (Mac) and ePSXe (crossover, winXP).
I keep hearing about patching clean ISOs and was wondering what are some of the glitches that you guys speak of? Perhaps you are trying to do something very extreme? If there are problems, has anyone managed to track down whether or not it is isolated to anything? Or rather, it is because of the way these patches work - some fundamental understanding that I am lacking of.
I don't plan to release my scrap-work patched ISOs so if it gets corrupted or something then so be it. But I need to be aware of the consequences from dirty patching for the future.
You cant get a dirty ISO (that i know of) from patching with FFTPatcher or Shi Shi, people get dirty ISO's from patching a PPF onto a ISO thats already been patched with a PPF then you will see problems like, delita knowing Ultima, some odd text problems such as chinese or blank, and some other stuff.
^ Thanks for that tip.
Now about those glitches, they seemed relatively rare to observe. What I am wondering is, are stuff like Delita knowing ultima suppose to be common? I have actually patched over 1.3 and 1.3 easytype (PPF patched ISOs), and well over a thousand times, so would the said problems actually suppose to show up somehow? Or are they just very specific problems? If these glitches are suppose to be rare, then I guess I just happen to have never seen them yet.
I haven't done much text editing for now, mostly because I am aware that text problems do arise (that I have seen). So I am not going to do text changes until everything else is near completion.
R999 isn't the only one. I repatch all my stuff too, with no consequence.
Here's the deal:
Patches contain changes. They compare a hacked iso with an unhacked one, and bytes that have been changed from the original are placed in the patch.
You can patch over a "dirty" ISO without problems IF little or no data movement has been done to the "dirty" one. Since the patcher largely just messes with data within tables, combining patches that just use changes from that shouldn't have too many problems, since you'd just be overwriting that data again. However, if ASM hacks have been applied or sectors moved in the dirty one, game destroying crashes become far more likely.
So pretty much, in an order of operations, you want to:
1. Patch
2. Event Edit
3. ASM Hack
.ppf
FFTPatcher & Shishi will crush all previous changes.
ppf files compare the difference between a clean ISO and the wanted patch. Huge difference.
So when something goes horribly wrong, it's the PPF patch that's likely the culprit.
@Kaijiyuu can you be more specific about byte movements? If I am getting this correctly, that means changes in Patcher and ShiShi(Sprite) would never move bytes around.
What I am interested in as well is the FFTactext patching. I wonder if bytes would be moved around in there? It should be very possible due to the way it works, unless all the strings are of fixed length (unlikely).
The older versions of TacText did not consider data compression. Those versions seriously messed up on the WORLD.BIN file. The newer version no longer has override errors because not only is data compression considered, if you exceed the maximum limit for a given file, it will also not allow you patch your ISO. Bytes shouldn't be moved around from file to file since each file has a maximum size reserved for text.
Quote@Kaijiyuu can you be more specific about byte movements? If I am getting this correctly, that means changes in Patcher and ShiShi(Sprite) would never move bytes around.
Moved? Probably not. Changed? Yes.
If the data at 0x08976F contains the dancer skillset (it doesn't, I just picking random numbers/stuff here), the data has been changed but not moved. If the dancer skillset is placed elsewhere, though, it would be moved and hence be likely to cause a crash.
TacText though very likely does move data and pointers around.