Hi there. I’m having a problem with Postgresql not being very happy with HoneySQL@v2 converting :foo/bar to "foo"."bar" when set-ing a new value. For example:

(-> (helpers/update :user)
      (helpers/set {:foo/bar 1})
      (helpers/where [:= :user/id id])
      (helpers/returning :*)
      (sql/format {:dialect :ansi}))

Open an issue on github and I'll fix it tomorrow. It's a bug.

Will do. Thanks 🙂

Hi there. Where should I be looking into if I want to configure automatic conversion between kebab-case and snake_case for both table names and column names? Thanks in advance.

it does that already

It doesn’t for me for some reason…

I’m using Hikari for pooling, and every time I use honeysql it always yell at me for not having the proper relation

are you using next jdbc?

yep

do you have csk on your classpath?

I dont know what that is sorry

camel snake kebab

oh yeah I do

I have it in lein

do I need to also require it like next.jdbc.datetime?

so here’s what I have atm:

;; conn
(connection/->pool HikariDataSource (select-keys env [:jdbcUrl]))

(s/defn read-user-tab :- [TabDB]
  [{:user/keys [id]} :- {:user/id s/Uuid}]
  (log/info "Reading tabs for user id" id)
  (nj/with-transaction [tx db]
    (nj/execute! tx (-> (helpers/select :*)
                        (helpers/from :tab)
                        (helpers/where [:= :tab/user-id id])
                        (sql/format {:dialect :ansi})))))

I’m getting yelled at that user-id relation doesn’t exist

it’s called user_id in the db

you don't need the :tab/user-id

just [:= :user-id id]

let me test real quick

@dharrigan No luck. It’s still giving me ERROR: column "user-id" of relation "tab" does not exist

with or without the tab namespace

changing it to user_id works as expected

question why dialect ansi?

because I want quoting

the quoting may disable the undasherizing

https://github.com/seancorfield/honeysql/blob/8ffca0deae44c6a1d546669cbcaa5c8bdb8baa32/src/honeysql/format.cljc#L75

@_@

holy heck

you’re right

I just removed it and it works

haha is that even intended?

probably. Sean wouldn't do anything without a reason 🙂

it's probably because as you're using ansi, it won't make any assumptions about the identifiers

it'll treat them "literally"

rather than trying to undasherize them

I mean I can rename the table, but it’s hard trying to find anything more suitable than user

I have another problem it seems. The returned map still keeps it as user_id…

Well that’s tomorrow’s problem. Time for me to do something else 😅

User is the only convention I break (keeping tables singular) and I call it users

makes life simple 🙂

@rextruong To clarify some stuff about the camel snake kebab stuff: that only affects next.jdbc and it only affects 1) names in ResultSet’s that the library converts to Clojure and 2) names in Clojure data structures that the library converts to SQL — it does not (cannot) affect anything in the SQL string itself.

HoneySQL V2 defaults to :ansi dialect so you don’t need to provide it as an option (it’s relatively harmless to do so). If you specify a dialect, you get quoting. You can also ask for :quoted true to get quoted with the current/default dialect.

If you ask HoneySQL to quote names, it does so without dealing with -/`_` — in both V1 (which is the code @dharrigan linked to) and V2 here: https://github.com/seancorfield/honeysql/blob/v2/src/honey/sql.cljc#L150-L169

user=> (-> {:select :* :from :tab :where [:= :tab/user-id 42]} (sql/format))
["SELECT * FROM tab WHERE tab.user_id = ?" 42]
user=> (-> {:select :* :from :tab :where [:= :tab/user-id 42]} (sql/format {:quoted true}))
["SELECT * FROM \"tab\" WHERE \"tab\".\"user-id\" = ?" 42]

See some of the important differences around V1/V2 handling of names here https://github.com/seancorfield/honeysql/blob/v2/doc/differences-from-1-x.md#option-changes