For information's sake, currently only the full screen hack relies on an arbitrary jump point. The graffiti hack is pretty standard, and will run on even a non-i300 (it just won't do anything). The dialer basically just passes a custom "command line" to the PhoneApp, so it isn't doing anything so horrible that it will break if/when the phone is updated.
As for the full screen hack, the source code I provided really shouldn't be used in the raw. A prudent developer would check the first few bytes of the jump-point, and maybe even the last few bytes of the function (debug symbol) to make sure they're there... if not, then don't call the jump point.
One other thing to note... My code is pretty nice in my opinion. I've looked at PDAapps stuff, and they seem to use a hard coded jump point address. Mine looks for the database resource and calls the jumppint as an offset from that. A side effect of this that it works with both UI and UJ rom version (Samsung hasn't changed the phExtn extension)... Who knows, though... this could change.
Also, for the more adventerous folks out there... this: I've pulled some relevant databases off the i300 and installed them in a hack version of the IIIc 3.5 rom. Guess what happens? Right, it crashes.

But before it does crash, it diesplays the statbar "Phone Off".. the area at the top where the battery meter is) Does this mean that its a standard 3.52 install with simply extensions done via installed applications and databases? Perhaps...
And given that we know that the phone does have a flash... we i'd say there's a good bet that its upgradable.
I'm about done. my 15 minutes of free time is up... but keep in mine this: I've been operating under the "no SDK ever" mentality all along. Using the methods I outline here should not be considered a bad practice... however developers should check the phone version and jump point before sending the phone off to try and execute a block of memory which might be a bitmap, notepad entry, or something more destructive.
Jake