It’s a shell problem, it has already been reported, so it’s just normal here, not out of abnormal
I can just add everything I have here (with _build/deps, every detail would be good)
SunPoke04
SunPoke04
Ok, not deps, there is too much there
It’s just mix phx.new template, I didn’t change anything
https://gitlab.com/SunP04/zaworld
1
filmor
Could you pretty please set the debug environment variables I suggested? Without this, everyone will just guess.
0
SunPoke04
Screenshot by Lightshot The rest is running ok, should be the same as in your local env.
Ran debug=1 mix_debug=1 mix phx.server
0
Eiji
The following code has been called:
https://github.com/elixir-lang/elixir/blob/v1.17.3/lib/mix/lib/mix/tasks/deps.compile.ex#L300-L313
I have also tried to recreate arguments passed to this function based on authors images and what I found on my machine:
dep = %Mix.Dep{
scm: Hex.SCM,
app: :telemetry,
requirement: "~> 0.4 or ~> 1.0",
status: {:ok, "1.3.0"},
opts: [
app_properties: [
description: ~c"Dynamic dispatching library for metrics and instrumentations",
vsn: ~c"1.3.0",
registered: [],
mod: {:telemetry_app, []},
applications: [:kernel, :stdlib],
env: [],
modules: [
:telemetry,
:telemetry_app,
:telemetry_handler_table,
:telemetry_sup,
:telemetry_test
],
licenses: [~c"Apache-2.0"],
links: [{~c"GitHub", ~c"https://github.com/beam-telemetry/telemetry"}],
doc: ~c"doc",
include_files: [~c"mix.exs"]
],
lock:
{:hex, :telemetry, "1.3.0",
"fedebbae410d715cf8e7062c96a1ef32ec22e764197f70cda73d82778d61e7a2", [:rebar3], [], "hexpm",
"7015fc8919dbe63764f4b4b87a95b7c0996bd539e0d499be6ec9d7f3875b79e6"},
env: :prod,
build: "c:/Users/pedro/Desktop/Coding/Elixir/zaworld/_build/dev/lib/telemetry",
dest: "c:/Users/pedro/Desktop/Coding/Elixir/zaworld/deps/telemetry",
hex: "telemetry",
repo: "hexpm",
optional: false
],
deps: [],
top_level: false,
extra: [
erl_opts: [:debug_info],
deps: [],
profiles: [
test: [
erl_opts: [:nowarn_export_all],
ct_opts: [ct_hooks: [:cth_surefire]],
cover_enabled: true,
cover_opts: [:verbose],
plugins: [:covertool],
covertool: [coverdata_files: [~c"ct.coverdata"]]
]
],
shell: [apps: [:telemetry]],
xref_checks: [
:undefined_function_calls,
:undefined_functions,
:locals_not_used,
:deprecated_function_calls,
:deprecated_functions
],
project_plugins: [:rebar3_hex, :rebar3_ex_doc],
hex: [doc: %{provider: :ex_doc}],
ex_doc: [
source_url: "https://github.com/beam-telemetry/telemetry",
extras: ["README.md", "CHANGELOG.md", "LICENSE", "NOTICE"],
main: "readme"
],
overrides: []
],
manager: :rebar3,
from: "c:/Users/pedro/Desktop/Coding/Elixir/zaworld/deps/bandit/mix.exs",
system_env: []
}
config = [
consolidate_protocols: false,
consolidation_path: "c:/Users/pedro/Desktop/Coding/Elixir/zaworld/_build/dev/lib/example/consolidated",
deps_path: "c:/Users/pedro/Desktop/Coding/Elixir/zaworld/deps",
deps_build_path: "c:/Users/pedro/Desktop/Coding/Elixir/zaworld/_build/dev",
lockfile: "c:/Users/pedro/Desktop/Coding/Elixir/zaworld/mix.lock",
build_embedded: false,
build_per_environment: true
]
command = ~s[escript.exe "c:/Users/pedro/.mix/elixir/1-17/rebar3" bare compile --paths c:/Users/pedro/Desktop/Coding/Elixir/zaworld/_build/dev/lib/*/ebin]
print_app? = false
env = [
{"REBAR_BARE_COMPILER_OUTPUT_DIR", "c:/Users/pedro/Desktop/Coding/Elixir/zaworld/_build/dev/lib/telemetry"},
{"REBAR_CONFIG", "c:/Users/pedro/Desktop/Coding/Elixir/zaworld/_build/dev/lib/telemetry/mix.rebar.config"},
{"REBAR_PROFILE", "prod"},
{"TERM", "dumb"}
]
Note: Pay attention that some of those values may be a bit different, but most of them if not all should be exactly the same as what author’s have.
As we can see the escript.exe return code is different than 0 which means it failed. However looks like there is not much info what exactly have failed.
@SunPoke04 Are you able to clone the Elixir repo, switch to v1.17.3, modify code as shown below, compile and run mix compile with updated code?
https://github.com/elixir-lang/elixir/blob/78f63d08313677a680868685701ae79a2459dcc1/lib/mix/lib/mix/shell.ex#L115-L134
Please wrap System.shell/2 call at the bottom of the function with IO.inspect/1 call, for example:
# …
{_, status} = IO.inspect(System.shell(command, options))
# or
{_, status} = System.shell(command, options) |> IO.inspect()
# or
{message, status} = System.shell(command, options)
IO.puts(message)
# …
As long as the code runs we do not care about if it’s nice or not. What we may be interested here is first tuple element i.e. error message coming from escript.exe which as you can see is unfortunately silenced.
Edit: Created a related issue:
https://github.com/elixir-lang/elixir/issues/13910
Edit 2: This is going to be even more interesting … Looks like there is no issue with printing output (as it’s done using :into option) which would mean that escript.exe have failed with non-zero exit status and did not print any message. Unless there is a way everyone can reproduce, it would be a very hard to debug.
I have created a post on Erlang Forums, see:
@SunPoke04 How about checking if escript.exe have all dependencies installed? In Linux this could be done with ldd command, but I have no idea about Windows. Consider checking topics like:
0
SunPoke04
@SunPoke04 Are you able to clone the
Elixirrepo, switch tov1.17.3, modify code as shown below, compile and runmix compilewith updated code?
I can’t compile it though, it seems I’m missing stuff, dunno. Screenshot by Lightshot
@SunPoke04 How about checking if
escript.exehave all dependencies installed? InLinuxthis could be done withlddcommand, but I have no idea aboutWindows. Consider checking topics like:
Screenshot by Lightshot it seems to work properly
0
Eiji
Heh, I never saw it, but it may be a huge hint!
Did you follow Windows wiki?
I guess making Elixir able to compile on your machine could be at least half a way to solve your issue.
1
SunPoke04
Did you follow Windows wiki?
Didn’t know there was one, ngl.
But I saw it now and it requires msys2, which erm… kinda meh.
I already have mingw stuff, but it also requires bash/grep, which isn’t that “normal” in windows.
I can install those, but I can’t promise I’ll do it since it’s not my style of doing stuff and I’ll probably just forget about it.
I have work+uni today and tomorrow, so if I am able to install msys2, I’ll do it on the weekend (sry for taking a lot of time in advance)
0
dimitarvp
You said you don’t want to use Docker, which is a valid choice, but what about WSL2? That’s practically how most Windows-bound programmers work these days.
2
SunPoke04
Sry, working has took a lot of my time, wasn’t able to do much outside of it.
I’m just not fond of WSL, I’ve used it in the past and it didn’t work properly, also it takes too much memory, so it’s a major pain
0