ci: add test for windows

[bjohansebas]

Related nodejs/citgm#1095 (comment)

[bjohansebas]

I discovered that the error is in this file

[UlisesGascon]

Maybe we can make it work with some tweaks:

#!/usr/bin/env bash

set -e

# Detect OS
OS="$(uname -s)"

# Define file paths
KEY_PATH="./test/fixtures/server.key"
CERT_PATH="./test/fixtures/server.crt"

# Generate the certificate
if [[ "$OS" == "Linux" || "$OS" == "Darwin" || "$OS" == "MINGW"* || "$OS" == "MSYS"* || "$OS" == "CYGWIN"* ]]; then
    echo "Running on OS: $OS"
    openssl req -x509 -nodes -newkey rsa:2048 \
    -keyout "$KEY_PATH" \
    -out "$CERT_PATH" \
    -days 3650 \
    -subj "/C=US/ST=Illinois/L=Chicago/O=node-express-session/CN=express-session.local"
else
    echo "Unsupported OS: $OS"
    exit 1
fi

echo "Certificate generated successfully at: $CERT_PATH"

So far seems like the -subj argument is the problem on the win environment (ref):

++ uname
+ OS=MINGW64_NT-10.0-20348
+ '[' MINGW64_NT-10.0-20348 = Linux ']'
+ '[' MINGW64_NT-10.0-20348 = Darwin ']'
+ openssl req -x509 -nodes -newkey rsa:2048 -keyout ./test/fixtures/server.key -out ./test/fixtures/server.crt -days 3650 -subj //C=US//ST=Illinois//L=Chicago//O=node-express-session//CN=express-session.local
......+..+.+..+...+....+.....+..................+......+....+..+.+..+...............+...+...+....+...+++++++++++++++++++++++++++++++++++++++*.+..+.......+.....+.+......+..............+......+++++++++++++++++++++++++++++++++++++++*..+..........+...+..............................+......+.....+....+........+.......+..+................+..+.+......+.....+.+..+...+...+......+......+..........+..+..................+....+...+........+..........+..+...............+.+......+.....+.+..+............+.+...+..+............................+.....+....+.....+...+........................+...+.+..+...+....+..+..........+.........+.........+.....+...+....+...........+....+......+..+.............+...............+..+...........................+.+...+.....++++++
.+.+........+......+.........+......+....+.........+..+...+................+.........+.....+......+....+...........+.+.....+.+...+..+...+............+...+.+......+++++++++++++++++++++++++++++++++++++++*..+....+..............+.+.....+....+.........+.....+.+..+.......+...........+....+...+.....+...+.+.........+...............+.....+.+...+..+................+........+.........+...+....+........+++++++++++++++++++++++++++++++++++++++*..+.....+.........+.+......+...+...+..+.............+............+.....+.+...............+.....+.+........+.+.....+....+..+..................+.+..+....+...+..+.+..+...+....+.....+....+..+........................................+...+...........+.+..+.....................+......+...+......+.+.....+.+...+..............+...............+.+.....+.+......+...+...............+.....+......+.........+................+.....+..................+.+..+....+...+.....+.+..+...............+....+.....+...+.......+..............+.+......++++++
-----
req warning: Skipping unknown subject name attribute "/C"
req warning: Skipping unknown subject name attribute "/ST"
req warning: Skipping unknown subject name attribute "/L"
req warning: Skipping unknown subject name attribute "/O"
req warning: Skipping unknown subject name attribute "/CN"

[raphendyr]

Would it be better to reverse this if? I mean, should we check that if the OS is Windows, then use the alternative command.

The CI currently tests in windows and linux, but BSD and other similar systems are supported as far as I know.

[raphendyr]

Easy switch case could be:

case "$(uname)" in
  MINGW*) ... ;;
  *) ... ;;
esac