That DX driver looks like a very good deal if you can live with the compromises in the design. I see them as:
1) 25v rated input cap. Maybe not the end of the world, but it would have cost them two cents to put a 50v cap on there instead, which would have given us a little more of a safety buffer in setups we're likely to encounter on reef tanks (that is, trying to drive more than 3 or 4 LEDs).
2) It has a bridge rectifier built in to the PCB. Most of us are used to supplying DC so this is not really required - but it could make for an interesting ultra-budget app if you wanted to just run it on an AC transformer as someone else mentioned above.
3) The product photos clearly show two different products, and some of the specs in DX's product listing don't match the chip's datasheet (input voltage range, etc.) which makes me nervous.
4) The DIM connection isn't brought out in an easy to use manner.
Also, some thoughts about the chip itself:
1) Dimming looks iffy low in it's range. People complain about the "shoulder" low down in the range of dimmable Meanwell ELNs all the time - this product looks like it would have the same issues.
2) It's not the most stable thing in the world, but again this probably isn't a big deal.
It would be neat to play with this chip "from scratch" and the SOT89-5 package is probably not bad to solder by hand (the other package has a pad on the bottom that's not exposed, so without reflowing, you'd need to use conductive epoxy to glue it on then solder the actual pins). But by the time you designed and built a driver, I dunno if it would be that much cheaper/better than what we've got in this thread so far - the DX version is more or less cheaper than a hobbyist could expect to DIY it from what I can tell (though I can't find a single vendor that carries the bare chip, so who knows). The only real advantage is that this is a switching reg, not a linear like the CAT4101, so you have a little more freedom in terms of differences in input/output voltage.