I appreciate your input, but I disagree. I find it useful to search for stories initially published after a given date, regardless of when it was last updated.

 

Fics go stale for lack of a better word.. I use it to pick a published date. Eg 2013 and then sort by review. This eliminates big stories before that. A quick and easy way to find popular stories.

 

A use case for updated before would be for stories (with HP for example) without certain book plot elements.

e.g want a story that has no chance of having horcruxes? Set the update before date for the day before DH was released.

There are uses for it, but they're more niche, which is fine.

Scryer exists to find stories, having more options to narrow searches is never a bad thing.

 

I've been rewriting how Scryer works under the hood, it's been reindexing the world and not focusing on keeping the existing index up to date. Should be solved soon.

 

The best way to do this currently is to use "Required Characters" (or Genres) to add characters you want, and Exclude Characters to remove the ones you don't want. e.g. Required[Harry P.], Exclude[Hermione G., Ron W.].

Include/exclude tags are somewhat as you'd expect. If you include a tag, all stories returned will have that tag, but may have other tags. Exclude allows you to then provide tags you do not want to be returned. e.g.

Story 1: tags [harem, slash]
Story 2: tags [harem, gender bender]
Story 3: tags [harem]

With a include=[harem] and exclude=[slash], Story 2 and Story 3 would be returned.

The tags are currently limited to system-generated tags, Scryer attempts to determine from characters, relationships, and the summary whether a story fits certain criteria for each tag. I do want to expose tagging to users in the future, but it's not high on my list given how few people would actually tag stories.

 

I've been replacing my Python logic which is terrible

 

After a lot of query rewriting to update from ElasticSearch 2.x to 5.x, Scryer is moved over. The index should be faster now. This also unblocks work on improving search features. As an added bonus, it's also up to date.

 

Was there really that big of a version difference? 2.x to 5.x seems like a big leap.

I'm excited to hear this.

 

ES 2.x is what was installed on Scryer's old index server. 5.x is what powers the DLP Forums. I wanted to unify the two. Unfortunately, there was no way to upgrade the old index and required reindexing all stories into the newer ElasticSearch, so I took the opportunity to fix some sins of the past.

2.x to 5.x isn't really a big change, though they removed a bunch of functionality. Mostly, they're going for the illusion of stability via larger version numbers to appeal to larger customers. 1.x was the version we used at my last job for a looong time. Then 2.x came out, and they skipped 3.x and 4.x and went right to 5.x.

 

Update: Sorting by DLP score works again.

 

Every day we stray further from God's light.