sendgrid-ruby | Official Twilio SendGrid Led , Community Driven Ruby API | Email library

Twilio SendGrid developer evangelist here.

Check out the documentation here that shows all the options you can use to send an email with the Ruby library (excuse the use of include).

Based on the docs, here's a quick snippet that should cover your needs:

Twilio SendGrid developer evangelist here.

The dynamic data that you are adding is more complicated than the template. Instead of sending an array of hashes as the dynamic data, you can just send a hash. Try this instead:

Okay, so after a lot of trial and error I finally got this working. I hope the code will be helpful to others.

So, firstly, what I did was send an actual event invite from iCalendar and receive this .ics invite (which actually got rendered in both Outlook and Gmail). I looked at how this file was different from what I was generating and found a curious thing:

the key to get this working was...

MAGIC STRINGS

Yeah, totally random, weird magic strings.

Below I'm posting the .ics file content that worked for me.

TOTTALLY-RANDOM-MAGIC-STRING - is a placeholder for a totally random strings like uuids or maybe your organisation emails or anything else.

The key is: with these strings in the file Outlook and Gmail render the invite correctly, and without them - don't. Weird, but working.

I wasn't able to find anything meaningful about this in the docs or RFCs, so I guess it's safe for now to call these magic strings.

The first magic string is TOTTALLY-RANDOM-MAGIC-STRING@imip.me.com.

And the second magic string is /TOTTALLY-RANDOM-MAGIC-STRING/principal/.

My issue was that a module was being included in the global namespace, and so its constants were polluting the entire codebase.

For example:

Gemfile.lock is a file generated from Gemfile. As such, instead of trying to merge the two branches, it's simpler and more accurate to generate a new one from its canonical source. This might result in slightly different versions, but these should cause no trouble; any version restrictions should be defined in your Gemfile.

Normally one does not commit generated files, they can change in trivial ways, but Gemfile.lock is a special case where you do want this to be the same for all builds.

Resolve any conflicts in the Gemfile. Regenerate Gemfile.lock. Add it.

for some reason he copied the code and added it himself

This is a good opportunity to explain to them why this is a bad practice when working with a team. It might be easy for them, but it's causing trouble for you. They might need instructing in how to update their work in progress. Or you might need to extract some changes into their own branch and get that merged.

Twilio developer evangelist here.

When you are sending emails with SendGrid using SMTP you need to authenticate yourself. SMTP uses basic authentication which requires a username and a password. In order to use an API key as the password, we also need a username.

Since the API key is all you need to actually authenticate yourself the username is a sort of placeholder. So, in the example the username is "apikey" but that is literally what it should be set to: "apikey".

The password should then be set to your API key, which you create in the SendGrid admin console. So the code should be something like:

You're mounting a named volume over the container's /usr/local/bundle directory. The named volume will get populated from the image, but only the very first time you run the container. After that the old contents of the named volume will take precedence over the content of the image: using a volume this way will cause Docker to completely ignore any changes you make in the Gemfile.

You should be able to delete that volumes: line from the docker-compose.yml file. I'm not clear what benefit you would get from keeping the installed gems in a named volume.

Thanks to @Les Nightingill, I found the issue.

It was not directly linked to assets generation, but the probleme was indicated at the first error line in the logs :

Generally one would use the async actioncable adapter when testing (not the redis adapter) and you'd need to configure actioncable to run in app (in the test environment) - https://guides.rubyonrails.org/action_cable_overview.html#in-app - in order for it to start up when Capybara starts the application.