Threadsafe, fast and clean fork - What to do?

[viblo]

I have rewritten large parts of the UdgerParser from the dev_v3 branch to make it threadsafe, faster and cleanup the code. It passes all the tests, and from my own testing is much faster than the previous version.

Compared to the previous implementation it caches most of the database in memory, except IP locations and IPv6 data center mainly because the IP location lookup table is huge. It adds a flag to do Data Center parsing (DataCenterParserEnabled) so that you enable data center parsing at the same time as IP location can be disabled (so you can run it fully in memory without disk reads and still parse IPv4 data centers).

We have been running it in production for a while, and it seems to be working well.

Currently its a fork available here: https://github.com/nordicfactory/udger-dotnet
I would be happy to make a PR if you think its appropriate.

[markatosi]

I'm very much looking forward to this. Thanks for your efforts I'll be happy to test.

[flo8]

Is this still alive?

[viblo]

We are still running this in production, processing many millions of user agents and IPs per day. But we haven't really made much changes since I created this issue, only fixed a couple of bugs that we encountered. Otherwise it has been stable.

[markatosi]

I have been using it briefly but with my data sets the full list scan being done is a performance bottle neck for us relative to the java version. I don't remember the exact name of the method but it has fullscan in it