[cogserver] ShellUTest sometimes fails (corrupted size vs. prev_size)

[Radivarig]

ShellUTest fails more often than it doesn't, so I disabled it.
The issue is already described here

[linas]

On my system, I can run this test 19x per hour, all of them pass.

[linas]

OK, so its a boost bug .. with boost 1.62 I find that 30 tests out of 30 pass.

With boost 1.67 I get 5 failures out of 24 runs...

Have to ponder how to get to a working version of boost ... since apparently older packages are not generally available ...

[Radivarig]

It's working with boost162 at least 5 times in a row that I've tested. It takes me ~200s to complete though. Am I missing some config like with atomspace LargeFlatUTest?

[linas]

I think there is a 200sec timer in there, because it always takes that long, no matter what. It's meant to take a long time, under the assumption that race conditions, if present, are rare.