# Org-roam: Technical Presentation
Leo Vivier

[[!template id=vid src="https://mirror.csclub.uwaterloo.ca/emacsconf/2020/emacsconf-2020--18-org-roam-technical-presentation--leo-vivier.webm" subtitles="/2020/subtitles/emacsconf-2020--18-org-roam-technical-presentation--leo-vivier.vtt"]]  
[Download compressed .webm video (23.5M)](https://mirror.csclub.uwaterloo.ca/emacsconf/2020/smaller/emacsconf-2020--18-org-roam-technical-presentation--leo-vivier--vp9-q56-video-original-audio.webm)

Org-roam is a Roam replica built on top of the all-powerful Org-mode.

Org-roam is a solution for effortless non-hierarchical note-taking
with Org-mode.  With Org-roam, notes flow naturally, making
note-taking fun and easy.  Org-roam should also work as a
plug-and-play solution for anyone already using Org-mode for their
personal wiki.

Org-roam aims to implement the core features of Roam, leveraging the
mature ecosystem around Org-mode where possible.  Eventually, we hope
to further introduce features enabled by the Emacs ecosystem.

The purpose of the talk is to present some technical aspects of
Org-roam.  From the very beginning, we wanted Org-roam to scale with
your notes, and this meant that we had to keep a close eye on our
performances.  As we iterated, optimisation remained a top-priority,
leading us to constantly peek under Org-mode's hood.  Not only has
this made us better developers, but it has also uncovered paths of
optimisation for Org-mode itself.

The talk is targeted at software engineers willing to peek under
Org-mode's hood.  A rudimentary understanding of Elisp will be
required.

Points to be covered

-   SQL database via emacsql
-   Elisp libraries
-   Parsing of Org-mode files
-   org-elements.e
-   Parsing with a background-process
-   Ensuring consistency via hooks

<!-- from the pad --->

- Actual start and end time (EST): Start: 2020-11-28T15.39.41; Q&A:
  2020-11-28T15.56.29; End: 2020-11-28T16.01.03

# Questions

## Why not to run a background Emacs for parsing instead of implement a new parser?
Running a background Emacs progress sounds great, but is still
limited. Forwarding all queries to a background Emacs (like org-mode's
exporter does) is only feasible with a (??? zaeph can probably fix the
answer).

## How often does the DB index get updated in order to contain changes within Org files?
Either on save, or on idle-timer.

## Did you ever think of opening up (or designing) the SQL DB as a general Org speedup-tool outside of org-roam so that other libraries that do execute complex queries are able to re-use the summarized data?
FYI, see John Kitchin's work, he uses a SQLite database to index his
Org
files. <https://kitchingroup.cheme.cmu.edu/blog/2017/01/03/Find-stuff-in-org-mode-anywhere/>.

- John's DB approach is great. However, we should not end up using
  several DB-index in parallel. ;-)+1

## Obviously with the 'global backlinks' agenda, it would be interesting to combine with the eev stuff from before :-) (<https://github.com/edrx/eev>)

## About the external program, you could just talk to the PANDOC guys (or Firn [Parses org-files into data structures with Orgize <https://github.com/PoiScript/orgize>], Logseq [OCaml & Angstrom, for the document parser <https://github.com/mldoc/mldoc>]), they're very helpful and have already a good org-mode parser

## Is it feasible to have this process of parsing org-roam following the LSP protocol? that would allow to be editor agnostic, and it would save the work to define the communication protocol and any other technical details.

# Notes
- "org-roam just wants to create backlinks".
- org-mode has many many files (377 lines in dired&#x2026; including
  .elc files).
- If you want to create an index of all the org files using the native
  format, it would be very slow. So org-roam uses a sqlite database.
- ripgrep (written in Rust) is more capable than grep; used by some
  Zettelkasten implementations.
- "Is there something we could do to import backlinks into Org mode?"
- "We've always tried to have an experimental ground where we can track backlinks"