MAILDEV_OUTGOING_SECURE is not honored

[jtudelag]

Hey!!

Awesome project!!

I just notice that no matter whether MAILDEV_OUTGOING_SECURE is set or not, it is not honored, defaults to unsecure always.
Code "compiled" from master.

bin/maildev 

MailDev outgoing SMTP Server smtp.com:25 (user:undefined, pass:undefined, secure:no)
MailDev webapp running at http://0.0.0.0:1080
MailDev SMTP Server running at 0.0.0.0:1025

env | grep -i MAILDEV
MAILDEV_OUTGOING_SECURE=true
MAILDEV_OUTGOING_HOST=smtp.com

Thanks!

[jtudelag]

Also, I investigated a bit further, and it seems that all the "booleans" input args are not being processed well when using env vars.

[bburky]

I'm not sure what the desired behavior is, but this change to options.js would allow the following to work:

FOO=false
FOO=true
FOO=bar
# Uses the default value
FOO=

diff --git a/lib/options.js b/lib/options.js
index 3e67bc9..f487eed 100644
--- a/lib/options.js
+++ b/lib/options.js
@@ -39,7 +39,11 @@ module.exports.appendOptions = function (program, options) {
     const flag = option[0]
     const envVariable = option[1]
     const description = option[2]
-    const defaultValue = process.env[envVariable] || option[3]
+    const envValue = process.env[envVariable] || ""
+    // If the env value is "true" or "false", parse it as a boolean
+    // Else, if it is a non-falsy value, use it
+    // Else, use the default value
+    const defaultValue = envValue.toLowerCase() == "false" ? false : envValue.toLowerCase() == "true" || envValue || option[3]
     const fn = option[4]
     if (fn) {
       return chain.option(flag, description, fn, defaultValue)

I can submit this as a PR, but I don't know if you want a more capable options parser that would also allow 0 and 1 which are often accepted in environment variables.

[bburky]

Commander may need to be updated to 3.0.0+ too, I didn't test this.
tj/commander.js#987

[reischlfranz]

I'm not sure what the desired behavior is, but this change to options.js would allow the following to work:
[...]

I can confirm that this change from @bburky fixes the issue with boolean environment variables. Can we have this code merged?

[dada051]

Don't know wich one of the solution is better than the other, but it should be great to have one quickly :)