It seems that it was a DNS/hosts problem, as I was using bb nrepl-server localhost:4444 and for some reason it takes too long to resolve localhost (I did not have it in /etc/hosts)

this seems to be happening to more people lately. it is a new trend to not have localhost in /etc/hosts?

I really don't know, and the funny thing is that everything else that i tried to reproduce this (listening with netcat or smth) just fires up instantly

I would really want to know what bb uses to resolve localhost and is so slow

https://github.com/babashka/babashka.nrepl/blob/edd3d613bfb9bf3adabfd0bda5c3f5c6ee85ec20/src/babashka/nrepl/server.clj#L31

I seem to be able to reliably reproduce this in a container doing this:

# Run any simple container
podman run --name debian-stable -it --rm debian:bookworm

> apt update && apt install vim curl
> curl -L -O 
> tar xzvf babashka-1.3.186-linux-amd64-static.tar.gz
> ./bb nrepl-server localhost:4444 # Starts instantly
> edit /etc/hosts and comment out localhost entry
> edit /etc/resolv.conf and comment out nameservers
> bb nrepl-server localhost:4444 # Slow start

None of the other programs have problem resolving localhost though, even instantly Tried dig, nslookup, netcat

what other programs

I told you, it's (.InetAddress/getByName host)

I'll do some digging to see how java's InetAddress handles this below

Thanks

alright

I kinda found out the cause, commenting in case anybody searches for the solution: If on a linux system where the localhost entry in /etc/hosts is missing (It really shouldn't) and you are using systemd-networkd that adds a search . option in /etc/resolv.conf then resolving localhost is gonna be slow for java, I really don't know how java handles the search option

maybe check the openjdk implementation

will do

although bb compiles from graalvm, is this openjdk behind the scenes?

graalvm is based on openjdk, they don't re-implement everything

although some things could be different / overriden for native-image. you can also check graalvm itself, it's open source

can you also try (http://java.net.InetAddress/getLocalHost)`

Nice poc! I'm reproducing the slowness using

❯ bb -e '(java.net.InetAddress/getByName "localhost")'
#object[java.net.Inet4Address 0x2a35a328 "localhost/127.0.0.1"]

In any way this is not babashka related, no need to respond to these anymore, it just triggered my curiosity

👍

The function below is the one getting partially applied. The problem is with :TRIGGERS/trigger-ddl. When I attempt to destructure using {:TRIGGERS/keys [... trigger-ddl]} or {trigger-ddl :TRIGGERS/trigger-ddl :as trigger} the pre-condition in (assert trigger-ddl) fails, however, if I retrieve using (:TRIGGERS/trigger-ddl trigger) it's all happy.

(defn trigger-attach-redef
  "Attaches redefinition statement to trigger metadata."
  [execution-id
   {:TRIGGERS/keys [TRIGGER_SCHEMA TRIGGER_NAME trigger-ddl]
    :as trigger}]
  {:pre [(assert trigger-ddl)]}
  (let [tool-ns (str *ns*)]
    (assoc trigger :migration/redef
           (->> ["/* {{tool-ns}}: TRIGGER Redef: {{TRIGGER_SCHEMA}}.{{TRIGGER_NAME}} */"
                 "DROP TRIGGER IF EXISTS `{{TRIGGER_SCHEMA}}`.`{{TRIGGER_NAME}}`;"
                 "DELIMITER $$"
                 "$$"
                 "{{trigger-ddl|safe}}"
                 "$$"
                 "DELIMITER ;"
                 ""
                 ""]
                (str/join \newline)
                <<))))

(->> {fetching collection of maps}
     ,,, ;; populating data used later, sorted map, s
     ,,, ;; fixing string content in a map value
     (map (partial trigger-attach-redef execution-id)) ;; <-- here
     (vec))

just taking a punt - :TRIGGERS/keys [TRIGGER_SCHEMA TRIGGER_NAME trigger-ddl] is what you have - what would happen if you did keys [:TRIGGERS/TRIGGER_SCHEMA :TRIGGERS/TRIGGER_NAME :TRIGGERS/trigger-ddl]`

😯 One can do that? I've never tried that way before. Let me try. The interesting thing is that I tried it in a clojure repl, same result, so it's really confusing. Will give feedback on this other style in a bit, thank you!

Negative. I'm dumbfounded. In this following code, I've added a println , the pre-condition check now doesn't have meaning. The code runs and prints out the expected values.

(defn trigger-attach-redef
  "Attaches redefinition statement to trigger metadata."
  [execution-id
   {:keys [:TRIGGERS/TRIGGER_SCHEMA :TRIGGERS/TRIGGER_NAME :TRIGGERS/ddl] :as trigger}]
  (println :execution-id execution-id :trigger-ddl ddl)
  {:pre [(assert ddl)]}
  (let [tool-ns (str *ns*)]
    (assoc trigger :migration/redef
           (->> ["/* {{tool-ns}}: TRIGGER Redef: {{TRIGGER_SCHEMA}}.{{TRIGGER_NAME}} */"
                 "DROP TRIGGER IF EXISTS `{{TRIGGER_SCHEMA}}`.`{{TRIGGER_NAME}}`;"
                 "DELIMITER $$"
                 "$$" "{{ddl|safe}}" "$$"
                 "DELIMITER ;" "" ""]
                (str/join \newline)
                <<))))

(comment

  (->> [{:TRIGGERS/TRIGGER_NAME "foo"
         :TRIGGERS/TRIGGER_SCHEMA "xyz"
         :TRIGGERS/ddl "<<<<<DDL-HERE>>>>>"}]
       (map (partial trigger-attach-redef 123))
       (map pprint)
       (doall))

  (clojure.pprint/pprint
   (trigger-attach-redef
    2345
    {:TRIGGERS/TRIGGER_NAME "foo"
     :TRIGGERS/TRIGGER_SCHEMA "xyz"
     :TRIGGERS/ddl "<<<<<DDL-HERE>>>>>"}))

  #_())

=>
:execution-id 2345 :trigger-ddl <<<<>>>>
{:TRIGGERS/TRIGGER_NAME "foo",
 :TRIGGERS/TRIGGER_SCHEMA "xyz",
 :TRIGGERS/ddl "<<<<>>>>",
 :migration/redef
 "/* mysql-migration.explore: TRIGGER Redef: xyz.foo * ....

Thanks for showing me something new today @U4C3ZU6KX 🙇 I couldn't find such an example in the Clojure docs. Unless it's this part:

(let [person {::p/name "Franklin", ::p/age 25}
      {:keys [::p/name ::p/age]} person]
  (println name "is" age))

This will serve the same purpose and is good enough for my needs.

(defn trigger-attach-redef
  "Attaches redefinition statement to trigger metadata."
  [execution-id
   {:TRIGGERS/keys [TRIGGER_SCHEMA TRIGGER_NAME ddl] :as trigger}]
  (assert ddl)
  (let [tool-ns (str *ns*)]
    (assoc trigger :migration/redef
           (->> ["/* {{tool-ns}}: TRIGGER Redef: {{TRIGGER_SCHEMA}}.{{TRIGGER_NAME}} */"
                 "DROP TRIGGER IF EXISTS `{{TRIGGER_SCHEMA}}`.`{{TRIGGER_NAME}}`;"
                 "DELIMITER $$"
                 "$$" "{{ddl|safe}}" "$$"
                 "DELIMITER ;" "" ""]
                (str/join \newline)
                <<))))

Think my issue was {:pre [(assert ddl)]}, with {:pre [ddl]} it's 👌

@U1M0BC7S7 That would make a good example. Maybe you could add a simplified version, for others who may haven’t seen it this way either.