Linas Vepštas

Presumably, its a boost::asio bug .. boost::asio has been a headache for 5+ years ... the easiest solution would be to re-write the code to just not use it ......

OK, as a boost asio replacement, there is just plain-old asio: in Debian, `libasio-dev` otherwise https://think-async.com/Asio/ .. but it seems like boost asio is just a copy/fork of this. Given...

If we instead assume that asio is not buggy, and that, instead, it is throwing some new kind of error that we are not yet handling, then `network/ServerSocket.cc` is the...

Well, boost asio is certainly still as buggy as ever. Debian-stable is still on boost-1.67 .... I'd like to see all uses of boost in opencog removed, but its a...

Well, in comment https://github.com/opencog/cogserver/issues/5#issuecomment-560156702 I put it into a loop and saw 5 crashes out of 22 runs. Nothing has changed since then ... I can try the test again....

Other intermittent test failures include the pattern-miner `SurprisingnessUTest` .. between that and the rule-engine, I have to hit the circle-ci "re-run unit tests" button 2-3-4 times before I can get...

See, for example, https://app.circleci.com/pipelines/github/opencog/atomspace/1722/workflows/28f80258-5816-4bfe-b4c2-7c992f45d3bf/jobs/19124

Yes, agreed that URE-based reduction is superior (in most ways, at least) -- its just that URE-based reduction is potentially quite difficult to implement, and risks being quite slow when...

> It's very much experimental for now Brand-new git repos are great places for highly experimental code. There's no need to promise that the code works or will ever work,...

Hmm. OK. The "few custom changes" should be fed upstream, where they can be audited and controlled by the package maintainers.