you'd think it'd be easy enough to do. i mean.. you can get a $2 usb wifi dongles now, how much could it possibly cost to do a serial to wifi link..
about $140. which really is not that much, but considering that i am probably going to break about 5 or 6 of these links before i iron things out.. uh.. that gets pretty expensive.
so for $140 you can get a wiport ($120), but you'll need an an antenna ($10) and a rs232 ttl to serial levels converter ($10) .
it really is tempting to spend some time putting effort into hacking a usb wifi dongle to work. because there are real linux drivers available for these you don't even have to do any reverse engineering. all of the info is right there in the code. here are the steps i would take.
1. install a fake universal usb driver so i can read/write whatever i want to the usb device.
2. strip down and get the linux driver talking from under the fake driver.
3. get a microcontroller talking usb.
4. compile the fake driver onto the micocontroller.
5. xoxo from the robot kids
the problem with all this is the simple project would end up turning into a big project. i have other more important things i should be spending my time on. getting caught up on this detail is silly.
so i still have not decided against the silly idea, but i do have an interim solution. i bought a wifi game adapter on buy.com for $33 shipped. now that means i just need to get my micocontroller talking ethernet. again.. you'd think that'd be cheap as in 5 cents. nope. the cheapest you can do is $25 from http://edtp.com . still $25 is a lot cheaper toy to break over and over than a $140 toy. the other nice thing is that since it is broke into two pieces.. ethernet and wifi adapter.. i can test to see if i broke the wifi adapter. i can even add a 3rd layer "surge protector" for a whole $2.. that of course would just be a crap router.
i did try some more ebay action and bought a 1996 intel printserver for $6 shipped. what the fuck is the difference between an ethernet printserver with a db9 rs232 on it and a serial to ethernet device server. i have no idea if it'll work, but i guess i'll find out. i don't really know what the hell a printserver actually does. it seems like it really is just a device server and the "print" part is just marketing? then why the hell do serial to ethernet device servers cost at least $100 when a printserver cost under $50? maybe it has to do with scope. by saying it is just a printserver you don't have to work with every possible serial device and don't even have to work about latency. though.. i did find some specs on a serial device server and it had complete shit latency. 29ms.. wtf. i can walk to california, eat dinner, and come back in 29ms!