The client is still in rush mode, but it's more of a controlled rush, and I'm mostly off the critical path now. Today was eclectic: making up a set of internal cables with correct wire lengths, assembling a sample unit, and then bringing up an SSH server (dropbear) on the gadget.
Turns out there were several things needed to get dropbear working (apart from a eucalyptus grove). First, libns*.so*, so that getpwnam would work (this was a case of missing shared libraries silently causing failures -- no error message, but standard library functions just don't work). Then suitably tailored passwd/shadow files. Then pty support in the kernel....
Oh, and those weird-sized screws? Turns out the captive nuts are #4-40 after all. The "#4-40" reference screw I'd grabbed to check them must have had munged threads.
Well, it seems to be working now. Tomorrow, more work on supporting the integration & testing effort. Saturday, a lazy day at the coast!
Also today, a first meeting with another prospective client, in need fo some mixed-signal cleverness. The project sounds interesting, though there doesn't seem to be a discrete chunk I could pull off and work on separately, so it'd really be more of a consulting project, rather than freelance engineering.
Comments