Fluff-wise, I would say Al-Qadim was the quintessential example of how to do a setting right. In short, everthing worked together pretty well, but it was easy to lift elements from it for other campaigns. A counter-example would be 1st ed Oriental Adventures, where the setting just felt like it was tacked on because, gee, I guess we should have some sort of setting for this. Also, Al-Qadim was described from the point of view of the folks who lived there, as opposed to how modern or medieval westerners would perceive it. I never played in Al-Qadim itself, but borrowed liberally from it for both my Greyhawk and homebrew campaigns.
As for crunch, I think they got a bit overambitious with the kit thing. Magi were more tied to the elements as opposed to schools, which seemed to work well. There were also Sha'ir, who got spells from genies on the fly. They were not bolt hurlers, as they needed time for their gen familiars to fetch the spells, but they were not limited to a certain number a day, so they were great for utility and long-duration buff spells. I liked the Kahin (kind of a mystic druid/shaman guy) and the Hakima (divine seer type). These three spellcasters were definitely sub-classes, where most of the rest of the kits were just a collection of minor bonus, penalties, and extra skills. I was a bit paranoid about the crunch at the time, as I picked up Al-Qadim right when TSR was starting to flood the market with odd rules variants for 2nd ed. Looking back, most of it was either quite good or fairly innocuous, which places it quite high on the scale.
