Index size question

[taobojlen]

Are there any general trends when it comes to the size of hound's index? For example, zoekt ends up with an index about 3x the size of the source code. What's the equivalent number for hound?

[yelizariev]

Hound keeps git folder to checkout for new version of the repository (vcs-*). Then hound generates new folder idx-*, which contains:

• zipped text files (folder raw/)
• trigram index (file tri)
• few small meta files

So, exact numbers depends on amount of binary files in the repos (such files are excluded from index).

For this repository https://github.com/odoo/odoo/tree/13.0/odoo the numbers are as following:

• vcs folder: 671 MB

• idx folder: 206 MB, including

  • tri: 45 MB
  • raw: 162 MB

I assume that the difference could came from the fact, that hound makes not a strict index and greps the files to confirm the findings.

FYI. I plan to add code viewer to the hound, which will add extra space usage:

• full repository history (to be able to show code at any revision)