Slashdot Mirror
Riot Breaks Out At Foxconn
Presto Vivace writes with news (as reported by Engadget) of a riot at Foxconn's Taiyuan plant, reportedly over guards beating up a worker, and writes "Something is going on at Foxconn. Do any Slashdotters know of a good source for news about Chinese labor disputes?" Reports of the riot are also at Reuters, TUAW, and CNBC, to name a few.
The robotic automated control systems are typically shit, but that doesn't mean it's not doable, it just means that mechanical and electrical engineers should not write robotic control systems, they should leave it to software engineers. In other words, it's the same problem that the Diamond Viper video cards had back in the day when they let EE's write the video BIOS, instead of hiring a software engineer to do it.
I recently spent some quality time programming a Toshiba CA10-M00 controller interfaced robot for the purposes of doing testing on capacitive touch devices, such as trackpads, and the programming interface at the lowest level was, to put it bluntly, incredibly badly designed. The one saving grace was "palletizing" mode, and all that let you do was do things like fill columns in a biological sample tray while moving the pallet on which it was situated over one row at a time, and then repeating the previous instruction.
In any case, the controller was pretty terrible, very limited in capability, and only capable of controlling 4 degrees of freedom without being ganged to another controller for the next 4 degrees of freedom; even then; you'd want to install optional interface modules to use for step-signalling between the controllers, rather than ganging them, based on the limited number of steps available under the control of a single controller, and the inability to do anything remotely useful in only 1000 steps (with 4 degrees of freedom, 1000 steps was pushing rationality as it was).
As delivered, the hardware didn't actually function (had to send it back once to have a servo replaced), and when driven from other than the EEPROM, the command language is insufficiently rich to perform motion on more than a single axis at a time (which basically meant writing a program to write a program, rather than controlling it directly). Additionally, the plat was oriented incorrectly, and there were no registration marks on any of the manual adjustments, and the robot was not set up to be capable of non-2-d self interference (read: if incorrectly programmed, it could beat itself to death).
To top all this joy off, they very much expected you to use a "teaching pendant" to do a single static program, and I had to reverse engineer how to talk to the thing with a documented list of serial functions, with no documentation of order or the requirements for baseline settings.
All in all, to get a suite of repeatable test motions that could be applied to multiple devices with different form-factors required some fairly clever hackery. What I ended up with was a library of code that could be used to write a program that could program the robot. The most interesting of those are not in the public repository, but the rest of the code is here: http://git.chromium.org/gitweb/?p=chromiumos/platform/touchbot.git;a=tree
The bottom line is that by using meta-programming, instead of using the default crap interface you get by applying teaching-pendant programming, it'd be pretty trivial to change over the location of a screw, or even radically alter the layout.
And just practically speaking, fetching a screw is a subroutine, putting in a screw is a subroutine, and where to put the screw in is a point in the X,Y,Z,R point table, if you wrote your code correctly in the first place, which you'd be unlikely to do if using the teaching pendant, but which was still technically possible using one. Which'd mean just rewriting the point table after issuing a region erase command to the robot controller over an RS232C link, after jamming the robot into a receptive mode with 5 other command would move the screw.
But doing the metaprogramming approach, it'd also be possible to radically alter the robot behaviour pretty trivially and be up and running on the real assembly line once you got your test line working correctly to the new model.
Which is to say, the argument that you can't as trivially recon
Yep, SuperKendall, BasilBrush, and a few others are there to try to eliminate adverse talk about Apple Corp.
I don't know why, but I assume they work for them. I can't really see why they'd die in a ditch defending Apple's relentless unethical and stupid decisions if they were only fans.
I guess there's more to Apple marketing than black turtle necks, rounded corners and misleading ads.
China should actually be the shining example of capitalism. A government that supports the creation of companies, while pretty much outlawing any and all labor unions and worker organisation, a bare minimum of worker protection (afaik you can't simply kill them if you don't like them anymore), no interference with your hiring, firing, paying or worker treatment policies...
And STILL we're not happy. What more could we possibly want them to do to be good capitalists?
We used to have a Bill of Rights. Now, with the rights gone, all we have left is the bill.