dynom / tysug Goto Github PK

I'm not liking YAML for configuring TySug, I'm currently considering the following formats instead:

• HCL -- JSON with comments
• TOML -- A more sensible YAML
• SAN -- HCL "improved?"

TOML

Looks a bit like INI and offers a good specification. However I find it a bit too complex for TySug's use-case.

Things I don't like:

• Certain notations aren't super obvious

Things I like:

• Clear documentation and specification
• Distinction between arrays and tables (hashmaps)
• Typed arrays (types can't be mixed)
• Native support for dates (With TZ's)
• Freedom of indentation (usage and tabs or spaces)
• Unique keys (keys can't be repeated)
• Large community adoption and seems battle-tested

HCL

Things I don't like:

• Poor documentation of specification
• Mixed type arrays

Things I like:

• Seems nice and simple
• Fair community adoption
• I like the syntax

SAN

Things I don't like:

• Not mature and limited test coverage
• Opinionated indentation (spaces only)

Thinks I like:

• Simple

Currently the lists are configured rather statically. One use-case could be that the lists are more dynamically, backed with a centralised storage layer (e.g. Postgres / Redis, etc.).

For SQL like systems it might be desirable to write a custom query, to produce a sorted reference list. Something along the lines of:

-- MSSQL
WITH domains (domain) as (
SELECT
  LOWER(SUBSTRING([email], CHARINDEX('@', [email])+1, 999)) as domain
  FROM [users]
  WHERE email LIKE '%@%'
)
SELECT
  domain
  FROM domains
  GROUP BY domain
  HAVING COUNT(domain) > 2
  ORDER BY COUNT(domain) DESC

However, in this particular example, the output probably doesn't change that much in a running system. So the results can be cached and refreshed once per day/week or so.

Support for Marking Typos.

Probably not particularly useful, but seems fun to implement.