For what it's worth, I just did a quick test. Basically, given the incorrect code where EN is incorrectly raised before establishing the output lines:

SETB EN
NOP
CLR RW
SETB RS
MOV P1,A
CLR EN

My question was how long the NOP delay could be extended before the LCD failed to honor the command. I went as far as going through a 16-bit timer overflow delay--65,536 instruction cycles, or about 0.071 seconds on the AT89S8252 @ 11.0592 MHz. The LCD still honored the command.

Obviously your mileage will vary based on the LCD you use, but at least on the LCD I have on hand it doesn't seem to matter whether or not you put the data on the line before or after you raise EN. The only thing that seems to matter is what is on the line at the moment EN is brought low. This probably explains why the previously incorrect code in the LCD tutorial worked when I wrote it and also probably explains why there wasn't much in the way of concerns expressed about the LCD tutorial for 7 years.

In any case, I've corrected the LCD tutorial to reflect the fact that the RS, RW, and data lines should be put on the bus before the EN line is brought high.

Regards,
Craig Steiner