Christopher J. Morrone

> This needs to take hostlist syntax in addition to or preferably instead of regex for identifying lists of aggregators or lists of producers. It's difficult to correctly use regex...

> To @morrone question, the idea is that all of the aggregators know about all the samplers, but they only actually communicate with the one's that have been started. I...

> Because all of the aggregators **know** about all of the samplers, but they are not responsible for them. So the start and stop are there to alert a given...

> @morrone, the purpose of the regex on the listen is to handle the case where we are monitoring multiple clusters. So we've got cluster A, B, and C. I...

@tom95858 How does this sound? I am thinking of adding a "empty_array_skips_row: true" field to the decomp static code. I am skimming the decomp code, and it looks like decomp...

I would really, _really_ like to see some changes to the approach taken here. First of all, I would love to see plugins only have to have implement known API...

I don't know how much further review I should continue to do. If my high level requests are addressed, I think a number of things will naturally fall out of...

What are the symantics of sample() in the new paradigm? Does sample() need to be reentrant, or will ldmsd ensure that it is not required?

> I think the structure extension is quite standard practice. I can't speak to whether it is common in some circles, but it certainly isn't _standard_. I think you would...

Temporarily, to move the conversation forward, I'll set aside the idea of exporting all the of the API functions in a plugin module (rather than exporting one function that returns...