Add mounted router / route to mounted location if no definition exists

[antoine-sachet]

A trailing slash breaks the endpoint path but is required when accessing a mounted endpoint. This can make the API inconsistent.

Consider:

# plumber.R
#* @get /root
function() {
  rnorm(1)
}

# sub_api.R
#* @get /
function() {
  rnorm(1)
}

# entrypoint.R
library("plumber")
root <- plumber::plumber$new("plumber.R")
sub_api <- plumber$new("sub_api.R")
root$mount("/sub", sub_api)
root

Running the API:

• url/root works
• url/root/ gives a 404
• url/sub gives a 404
• url/sub/ works

Is there a way to make /sub be equivalent to /sub/ and /root/ to /root? Or is it a bad idea?

[schloerke]

Are trailing slashes required when visiting regular endpoints?

[antoine-sachet]

No, you're right, it's mostly url/sub not working which is very annoying as it creates an inconsistent experience for the API user because of an implementation detail.

It would be nice to have /root/ work as well (to make life easier for the API user), but this is completely secondary and, if it makes life harder for the programmer, it is not worth it.

[schloerke]

I'm in the camp that trailing slashes matter and should be treated as separate routes as they are separate routes.

I also understand the ease of use and common pattern of defining the route once and having it work in both places. I don't think I can enable it by default at this point, but adding a method (or init flag) would work well and tell of route mismatches.

[schloerke]

Closing in favor of #100

[antoine-sachet]

I agree with you, the routes are different and should be treated differently.

However, my point is different from issue #100 and I disagree with the renaming of this thread to "Trailing slashes should route to the same place".

My core issue is that when you mount an endpoint on /endpoint, then /endpoint does not actually work. You have to call the mounted endpoint with a trailing slash (/endpoint/) and this is very inconsistent, since trailing slashes are not otherwise required.

I maintain a relatively complex API whose main components are broken down in several plumber files and mounted (to e.g. /report, /review, /user)

People who integrate with our API ask me why they must put a trailing slash to /report/ but mustn't put one for all the endpoints below /report/ (like /report/<id> or /report/<id>/reset). Same thing for /review/, /user/, etc.

As far as the end-user is concerned, it seems like some random endpoints take a trailing slash while most others do not. Feels like a leaky abstraction?

[schloerke]

Are you wanting the example above to work like the lists below?

• url/root works
• url/root/ gives a 404
• url/sub gives a 404 works
• url/sub/ works gives a 404

And if sub_api had an extra route...

• url/sub/extra works

---

The way it currently implemented, you are allowed to have a custom route at url/sub with a mounted router starting at url/sub/.

So, would it be good to add the / mounted route to the base location? So it would be

• url/root works
• url/root/ gives a 404
• url/sub works ; (defined as url/sub/ if url/sub is not already defined)
• url/sub/ works

[antoine-sachet]

Exactly! Sorry I wasn't clear in my initial post, I tried to make it generic but just confounded the issues.

[schloerke]

Notes: Should be done at the same time as trailing slashes

[blairj09]

This could be accomplished if there was a way to opt-in to redirects on mounted routers:

options(plumber.mount.redirect = TRUE)

or

pr_mount(..., redirect = TRUE)

Example redirect:

pr_get("/foo", function(res) {
      res$status <- 307
      res$setHeader(name = "Location",
                    value = "/foo/")
      res
    })