Create new Java project includes a default http trigger function

[ejizba]

The file 'Function.java' should not exist and the user should just create a trigger themselves

As far as I can tell, this is also the only reason we have to prompt for 'packageName' when creating a project

[jdneo]

This can aligin with the js function project creation, but what I am concerning is that Java developers might feel strange if they find there is no source file in the maven archetype.

[ejizba]

The background implementation of a feature should not affect the experience of that feature. VS Code users don't know or care that you're using the maven archetype in the background

[akaroml]

@EricJizbaMSFT Why is a project with no code useful to the users? Can you elaborate or provide some success cases?

[ejizba]

I agree a project with no code is not helpful. However, my problem with the existing experience is that I have no control over the default trigger that is created. In an ideal scenario the user creates a project and is immediately prompted for the first trigger, including:

1. function name
2. type (http, timer, blob, queue, etc.)
3. settings (like AuthorizationLevel for an http trigger)

The default Function.java is helpful for brand new users, but is likely 'throw-away' code for a user with any experience. If the user can configure the first trigger - both new and experienced users should be happy

[akaroml]

Valid point, thanks. And two follow-ups:

1. Deleting code for experienced users shouldn't be much trouble. But finding a clue to what's next would be a blocker for beginners. We don't have many experienced users for Azure Functions at this stage.
2. Asking 3 more questions when creating a project in the command palette will make the process too long and too much.

I would recommend revisiting this issue when we decide to introduce graphical UI. Form is the right way to go when multiple entries need to be filled. For now, let's put this issue in backlog.

[ejizba]

For 1, it's not about how difficult it is to delete code, it's about how annoying it is to delete code.

The biggest reason I'm still debating this is because the Java experience is not consistent with other languages, which don't include code by default. If we're not going to be consistent, we should do it because the experience is without-a-doubt better (the current experience is debatabely better).

We initially thought multiple prompts would confuse users (as you hypothesized), but we did customer interviews and found it was not a big problem. I don't see VS Code moving to graphical UI anytime soon and I don't think that's a reason to wait.

[ejizba]

Moving to next release