Hi Jonas, thank you, your two suggestions are fair. I have had these in mind, and I will implement at least the first one in the next part of this blog post (when I get more data). This post a bit rushed, as I noted in the beginning.
As for the stacked graph for the breakdown, I wanted to do that too, but could not find a…
> The answer is straightforward: multiple columns simply mean that there’s more than one polynomial to work with
If there are multiple columns, doesn’t it means that the polynomials will be multivariate? With one column, you have f(x), with two columns, f(x,y), with tree columns, f(x,y,z), etc.?
You mention being grown-up and tantrums, yet a threat of not participating does seem like a childish tantrum.
There was no threat of not participating. I wrote that I will continue to provide updates for the calls and participate elsewhere. My concern is that being on a public call like this, without any rules to protect…
Firstly, it does not store information like Transaction hashes and block hashes with every single log entry. Secondly, it does not store bloom filter (256 bytes) with every transaction, but calculates it on the fly from the logs during the RPC request.
For this test, I was using a virtual server from Google cloud, with 4 CPUs, 1024Gb standard SSD and 120Gb memory. It would run on a laptop, but it will be slower. I will be publishing results for lower memory machines and HDD once I have stabilised the code a bit more
I do not have any such estimates yet, but that’s one of the goals. I am currently at the stage where I need to finish the Proof Of Concept at least (which means trying to impement all the changes). Before I am comfortable really discussing it, I also need to get some comparative performance numbers.