That is correct... and true, that Canon wisely choose to block settings where the "mix" (aperture, shutter, etc) will result that most of the buyers will get satisfying results on their daily use (Prio 1: Sharpness (of the object also when "moving"), Prio 2: Exposure/Colors etc...).
It's similar to "chip-tuning" a car: You will get with the same hardware (normally) more power (and some people report even of less fuel consumption but higher tire wear) but the default settings of car companies are wisely set to keep the balance between durability, (geographical) usability on fuel, temperature, air pressure (heigth) and "drivers".
CHDK doesn't make the pictures better by default....but used in manual mode (including background knowledge of photography) it enlarges the limits of the compacts significantly (keyword "creativity")... and technically it is interesting to see that the "cheap" compacts are actually capable of "much more".
So regarding diffractions: yes this may very well happen in some situations....but as seen above: there is an very acceptable "use" of the override :-)
Side-discussion: I'm not sure that the so called "aperture" setting is still "physical" (up to F11 mostprobably yes ...seems also to be as "text" within original firmware)... but at the high rates of F18 or F33, I think not.
If i'm right, I once read once an article of how - based on pure "maths" - things like aperture/shutter/iso etc, are handled on Digicams: Was something like that that there is a "dynamic" range of sensitivity of the CCD/CMOS. The range can be shifted as well the range "squeezed". Now let's assume that we can decrease the overall sensitivity of the CCD/CMOS chip to need a higher quantum of light until pixels are generating a value --> would result into same "feeling": Darker picture and need of longer shutter speed. Now because the CCD/CMOS needs "longer" to set the values, this may have the positive impact onto "precision" of the pixels --> "pseudo-high-aperture" :-)
ISO setting is a similar "virtualisation"...and could be based on the "time" the CCD/CMOS Pixels gets light (High ISO = put CCD/CMOS onto highest sensitivity, process the values as they come in, Low ISO = take more time so that statistically each Pixel gets more precise values).
So "fringing" (which is normally an optical/lens issue) would be not so much of an issue, because optically we are at max F11...
Eventually somebody (with real insight) may confirm...
