Always prepend -m with -M to avoid deprecation warning

[rrudakov]

Hopefully close #37

[vemv]

Thanks much! Looks nice to me, will take another look asap

[vemv]

@rrudakov my analysis would be that this namespace should generally forward what's given as an argument - it shouldn't fix stuff for the user.

Our official clojure.sh script already uses -M -m.

I suspect that what happens is that Elisp side, defun cider-clojure-cli-jack-in-dependencies does not emit a correct -M -m call?

Analysis / PR welcome for cider.

Cheers - V

[rrudakov]

Hi @vemv. This PR is not related to elisp or emacs. I want to start nrepl outside of emacs and connect to it using cider-connect (I followed this manual). The problem is that generated command in .enrich-classpath-deps-repl doesn't have -M before -m, so I'm constantly getting a deprecation warning when I'm trying to connect to nREPL.

[rrudakov]

A bit more details.

I have the following alias in ~/.clojure/deps.edn:

:nREPL
  {:extra-deps {com.kohlschutter.junixsocket/junixsocket-common        {:mvn/version "2.9.0"}
                com.kohlschutter.junixsocket/junixsocket-native-common {:mvn/version "2.9.0"}
                nrepl/nrepl                                            {:mvn/version "1.1.1"}
                cider/cider-nrepl                                      {:mvn/version "0.45.0"}
                refactor-nrepl/refactor-nrepl                          {:mvn/version "3.9.1"}}
   :main-opts  ["-m" "nrepl.cmdline"
                "--middleware" "[refactor-nrepl.middleware/wrap-refactor,cider.nrepl/cider-middleware]"
                "--socket" "nrepl.sock"]}

If I copy the Makefile from the README and add :nREPL alias to DEPS_MAIN_OPTS the generated command in the .enrich-classpath-deps-repl will be a huge string which ends with J-Djdk.attach.allowAttachSelf -m nrepl.cmdline --middleware [refactor-nrepl.middleware/wrap-refactor,cider.nrepl/cider-middleware] --socket nrepl.sock.

As you can see -m is not prepended with -M, but it should be.

clojure.sh adds main-opts in a different way as far as I can see, if I jack-in from emacs the command is the following:

/Users/rrudakov/.emacs.d/elpa/cider-20240220.720/clojure.sh /opt/homebrew/bin/clojure -Sdeps \{\:deps\ \{nrepl/nrepl\ \{\:mvn/version\ \"1.1.1\"\}\ cider/cider-nrepl\ \{\:mvn/version\ \"0.45.0\"\}\ refactor-nrepl/refactor-nrepl\ \{\:mvn/version\ \"3.9.1\"\}\}\ \:aliases\ \{\:cider/nrepl\ \{\:main-opts\ \[\"-m\"\ \"nrepl.cmdline\"\ \"--middleware\"\ \"\[refactor-nrepl.middleware/wrap-refactor\,cider.nrepl/cider-middleware\]\"\]\}\}\} -M:cider/nrepl

It doesn't pass main-opts to clojure, it just uses alias, not sure if the same effect could be achieved with Makefile.

[vemv]

Would it be valid if deps.edn specified the following?

:main-opts  ["-M" "-m" "nrepl.cmdline"

(edit: probably this is a bad idea if not invalid - better focus on my next post)

[vemv]

Also, the Makefile says:

enrich-classpath/examples/Makefile

Lines 28 to 29 in f3244ba

	# Expected format: "-M:alias1:alias2"
	DEPS_MAIN_OPTS ?= "-M:dev:test"

So -M is expected - are you passing it? Does the code strip it later?

[rrudakov]

vemv ***@***.***> writes:

Would it be valid if deps.edn specified the following? ``` :main-opts ["-M" "-m" "nrepl.cmdline" ```

No, I've tried. In this case I cannot use this alias as `clojure -M:nREPL` normally from command line. -- Best regards, Roman

[rrudakov]

So -M is expected - are you passing it? Does the code strip it later?

Yes.

My Makefile:

.PHONY: enrich-nrepl nrepl
HOME=$(shell echo $$HOME)
HERE=$(shell echo $$PWD)

.DEFAULT_GOAL := nrepl

SHELL = /bin/bash -Eeu

DEPS_MAIN_OPTS ?= "-M:local/backend:test/clj:nREPL"

ENRICH_CLASSPATH_VERSION="1.19.0"

.enrich-classpath-deps-repl: Makefile deps.edn $(wildcard $(HOME)/.clojure/deps.edn) $(wildcard $(XDG_CONFIG_HOME)/.clojure/deps.edn)
	cd $$(mktemp -d -t enrich-classpath.XXXXXX); clojure -Sforce -Srepro -J-XX:-OmitStackTraceInFastThrow -J-Dclojure.main.report=stderr -Sdeps '{:deps {mx.cider/tools.deps.enrich-classpath {:mvn/version $(ENRICH_CLASSPATH_VERSION)}}}' -M -m cider.enrich-classpath.clojure "clojure" "$(HERE)" "true" "-M" $(DEPS_MAIN_OPTS) | grep "^clojure" > $(HERE)/$@

enrich-nrepl: .enrich-classpath-deps-repl
	@if grep --silent "^clojure" .enrich-classpath-deps-repl; then \
		eval $$(cat .enrich-classpath-deps-repl); \
	else \
		echo "Falling back to Clojure repl... (you can avoid further falling back by removing .enrich-classpath-deps-repl)"; \
		clojure $(DEPS_MAIN_OPTS); \
	fi

nrepl:
	clojure -M:local/backend:test/clj:nREPL

the full generated command in .enrich-classpath-deps-repl is:

clojure "-M:local/backend:test/clj:nREPL" \
-Sforce \
-Srepro \
-J-XX:-OmitStackTraceInFastThrow \
-J-Dclojure.main.report=stderr \
-Scp src/clj:test/clj:test/resources:env/dev/clj:resources:/Users/rrudakov/.m2/repository/org/clojure/clojure/1.11.1/clojure-1.11.1.jar:/Users/rrudakov/.m2/repository/org/clojure/core.async/1.6.673/core.async-1.6.673.jar:................huge-list-here...........................:/Users/rrudakov/.m2/repository/io/netty/incubator/netty-incubator-transport-native-io_uring/0.0.18.Final/netty-incubator-transport-native-io_uring-0.0.18.Final-linux-x86_64.jar:/Users/rrudakov/.cache/mx.cider/enrich-classpath/17010/3847723509/582121401.jar:/Users/rrudakov/.cache/mx.cider/unzipped-jdk-sources/17010 \
-J--add-opens=jdk.compiler/com.sun.tools.javac.tree=ALL-UNNAMED \
-J--add-opens=jdk.compiler/com.sun.tools.javac.code=ALL-UNNAMED \
-J-XX:-OmitStackTraceInFastThrow \
-J-client \
-J-XX:+TieredCompilation \
-J-XX:TieredStopAtLevel=1 \
-J-Xmx4g \
-J-Dclojure.tools.logging.factory=clojure.tools.logging.impl/slf4j-factory \
-J-Djdk.attach.allowAttachSelf \
-m nrepl.cmdline \
--middleware [refactor-nrepl.middleware/wrap-refactor,cider.nrepl/cider-middleware] \
--socket nrepl.sock

I see my aliases are specified right after clojure. Actually I'm not sure that we need to pass :main-opts at the end at all, won't they be picked from the :nREPL alias?

[vemv]

Actually I'm not sure that we need to pass :main-opts at the end at all, won't they be picked from the :nREPL alias?

Probably it came to be like that for a good reason (cannot immediately check).

Thanks much for helping out with the analysis.

I'll give it another review during the week.

Perhaps you can contribute another unit test for the -A behavior? If the user specifies -A, should a -M be prepended? Maybe the right behavior is different for -A relative to -M?

[rrudakov]

vemv ***@***.***> writes:

I'll give it another review during the week.

Sounds good.

Perhaps you can contribute another unit test for the -A behavior? If the user specifies -A, should a -M be prepended? Maybe the right behavior is different for -A relative to -M?

Good point, I'll try to do some research on the topic a bit later. -- Best regards, Roman