This is partly to ask if anyone knows what's happening, and partly to offer a workaround to anyone in the same situation.
I have two eZ430-CHRONOS units, with their two MSP-eZ430U USB programmers, and as it happens, two computers to run on: a Dell PC running WinXP, and a MacBook running Windows 7 under Parallels. All combinations of all the hardware show this same behavior.
When I click the "debug" button in IAR --- which is a full version with dongle-locked license --- after a few seconds, the watch screen clears, but then IAR hangs (sometimes I get a "Not Responding" from Windows, sometimes a wait icon). Here's the workaround: if I unplug the dongle with the watch on it, after a few seconds I get
Emulator
Communication error.
Please reconnect the device and press Retry to reconnect, or press Cancel to abort
I then plug the dongle and watch back in, wait ~5 seconds, and click Retry. At that point the program gets downloaded --- though it takes probably a solid minute --- and debugging proceeds normally. If I click Retry too fast (presumably too fast to let the OS fully absorb the newly-re-plugged-in device), I get a "Failed to initialize device!" and have to start over.
Because this is behavior that is effected (and affected) when the USB dongle is unplugged and replugged, I am guessing it is a USB driver issue, which is why I'm posting here to the TI forum first. I am using the 21 August 2009 versions of hil.dll and msp430.dll --- if I use the older versions as in http://e2e.ti.com/support/microcontrollers/msp43016-bit_ultra-low_power_mcus/f/166/t/18596.aspx , I get different failure messages, they come faster, and the workaround doesn't work. This again suggests the problem's source is closely related to the USB drivers.
The fact that two watches, two watch-programming dongles, and two completely different machines running two (arguably three) different operating systems, all behave identically, suggests that this isn't an isolated incompatibility issue. On the other hand, I can't find any mention of this type of problem, and am hoping I'm just missing something obvious, at which point I will apologize profusely for wasting people's time.
On the other other hand, if there are people frustrated out there because they can't even start a debug session, well, hopefully the unplug/replug/wait/retry trick will get you hobbling along until a real solution shows up.
Thanks to all.