Ben Trask
Ben Trask
https://news.ycombinator.com/item?id=10497164 http://lkml.iu.edu/hypermail/linux/kernel/1510.3/02866.html Gotta be careful. It's very easy to make things gratuitously complex without really improving security.
There are basically two ways to do this: 1. By applying a cost factor to every step the filter does internally 2. By setting a timer I'm not sure which...
We need an API for loading several meta-data attributes at once. Then we should load all of the meta-data for all of the results in one transaction. We know the...
I don't know much about how to do this... Something like https://en.wikipedia.org/wiki/HyperLogLog ?
Sequential uses the system image libraries, so if they support WebM, it should be very easy to add. If not, you would need to integrate a decoder library, which would...
I doubled checked and WebM isn't natively supported in 10.10 (which you probably already knew). That means adding support would be harder, but still not impossible. There's also https://github.com/btrask/Sequential-3, which...
Rather than GPG I'm thinking [OpenBSD's Signify](http://www.openbsd.org/papers/bsdcan-signify.html) would be the best tool for this.
Alright, thanks.
Considering using the varint format from SQLite4: https://www.sqlite.org/src4/doc/trunk/www/varint.wiki Our current varint format can only pack values 0-15 into one byte. This hits us on every key's table ID.
A problem with the above idea of using sequential IDs instead of hashes for internal storage is that we can end up with junk in the file system when a...