hound-search/hound

Treating a single "repo" as multiple

Open

#172 opened on Nov 18, 2015

View on GitHub
 (3 comments) (0 reactions) (0 assignees)JavaScript (5,470 stars) (606 forks)batch import
enhancementhelp wanted

Description

Hi there!

I'm attempting to set up Hound to search the code from every WordPress plugin, which is approximately 48.5k separate repositories. I'm using #156 to avoid needing to have actual repositories for these, and instead using a home grown tool; this avoids needing the overhead of svn/git history. I'm then using an external Python script to generate the config.json

(Per #138, it seems like it might be copying the data across too? This is something like 20GB worth of data in the repos, so copying it is a tangible cost that I'd prefer to avoid, but one step at a time ;) )

From what I can see, it appears that there's significant overhead from having each of these repos separately; it appears hound is generating an index for each repo? (It also runs into #139 despite bumping ulimit -n significantly.) They're all quite small repos, so ideally, I'd like to have them in a combined index instead.

With a single combined index however, I can't build the URLs how I'd like. My directory structure is /srv/sync/plugins/<project>/<files>, but the URL structure is .../browser/<project>/trunk/<file>. With separated repos in the config, I can get this correctly, but not with a single combined repo.

I'm not sure how best to solve this in a way that minimises duplication but still gives me the flexibility I need. Any immediate thoughts on how to attack this? Also, any thoughts on how to run Hound at this sort of scale would be appreciated. :)

(Happy to contribute code as needed. The motivation behind this is basically that whenever we want to make potentially breaking changes in WordPress, we want to check against the actual uses in the wild so we can determine impact.)

Contributor guide

Tech stack
None
Domain
performancetooling
Issue type
feature
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
4
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
over 1 week
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
stale
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
mostly clear
Prerequisites
None
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
20
Research direction
The user wants to combine multiple small repositories into a single index to reduce overhead, while maintaining correct URL mapping based on directory structure. The issue suggests modifying Hound's indexing to treat a directory of repos as a combined index with custom URL patterns. Investigate the code in 'hound/indexer' and 'hound/hound' for how indexes are built and how URLs are generated. Consider whether a configuration option to specify a combined index with per repo URL prefix could be added.