On Saturday, 15 August 2015 at 08:28:59 UTC, wobbles wrote:
Nice work! Load times are excellent.
I've been meaning to do something like this for ages, never got
I'll just use this now. How open to pull requests / alterations
are you? I might try to do up a comment system for it.
Sure, as long as it fits general idea / goals. Comment system is
welcome but it can be quite challenging within existing design.
Right now I am using immutable cache for all data which gets
completely rebuilt (and saved to disk) as soon as any new article
is added. Which allows for cheap parallel access to data from
multiple worker threads without any synchronization and doesn't
harm performance as adding articles is rare and only done by blog
For comments that generally won't work - adding those can happen
often and is up to users. I think best approach would be to have
dedicated thread that owns all comment data and have worker
threads communicate with it via message passing - and make it
generally independent from immutable storage. Also save data on
disk periodically and not immediately.
Or maybe just say that this pushes NIH syndrome too much and
require external database to enable comment support. Not sure.
This is pretty much why I haven't implemented it yet :)