Infer var metadata for 'proxy' vars #171
Conversation
| @@ -41,7 +42,7 @@ | |||
| (defn- maybe-add-spec | |||
| "If the var `v` has a spec has associated with it, assoc that into meta-map. | |||
| The spec is formatted to avoid processing in the client (e.g. CIDER)." | |||
| [v meta-map] | |||
|
Is |
|
I can see how 'proxy' is an overloaded term, but 'aliased' immediately brings to mind namespace aliases, which would seem greatly confusing to me (particularly in a project that already deals with aliases).
|
|
When I hear of Clojure and proxy I immediately think of https://clojuredocs.org/clojure.core/proxy That's why I don't like this. And your examples in the tests are really aliases, after all this terminology extends to more than namespaces. I'm open to other suggestions, even if I can't offer anything better than |
|
Btw, I don't want us to spend too much time on this - the implementation seems good and solves a real problem. If we don't come up with a better name relatively quickly (a day or so), I'm fine to stick with the current terminology. |
| (:require | ||
| [orchard.misc :as misc]) | ||
| (:refer-clojure :exclude [find-ns find-var all-ns ns-aliases])) | ||
| [clojure.string :as string] |
There was a problem hiding this comment.
It seems that str is the preferred alias across the project, any reason to have it different here?
There was a problem hiding this comment.
Agreed - it's best to be consistent about this.
There was a problem hiding this comment.
If you don't mind I'd like to settle onstring then, project-wide. We can add a kondo rule for avoiding further churn.
My point of view as someone who tends to maintain other people's code a lot, is that Sierra-style aliases only make our lives easier.
There was a problem hiding this comment.
It's not a big deal for me, but such changes should be discussed outside the scope of this PR.
There was a problem hiding this comment.
I see str much more often than string (almost never the latter, actually). I don't know about feverishly enforcing some arbitrary rule one person established. Besides, he mentions exceptions in his blog post too.
There was a problem hiding this comment.
For me, most of all it's about not memorizing arbitrary abbreviations someone wrote years ago e.g. cljs.analyzer.api :as ana - those represent unnecessary cognitive load to me.
I agree that str is very well-known, but when one uses a convention 99.9% of the time, having an exception to break the rule isn't exactly convenient.
There was a problem hiding this comment.
So then everyone will have to memorize whether this is a project that chose string because of a convention or, you know, the rest 99.9% of the projects 😄.
Having said that, I claim no say in this issue.
src/orchard/info.clj
Outdated
| @@ -98,37 +98,47 @@ | |||
| ;; a special symbol - always use :unqualified-sym | |||
| (some-> (cljs-ana/special-meta env unqualified-sym) | |||
| (cljs-meta/normalize-var-meta)) | |||
| ;; an NS | |||
|
|
|||
| ;; a NS | |||
There was a problem hiding this comment.
Well ackshually, an NS is correct.
There was a problem hiding this comment.
Indeed, it's pronounced as en es, so the normal rules about consonants don't apply here.
I absolutely agree; I was reading through the code waiting and waiting where a I think it's worth picking up a better term. |
|
Thanks for the input! I'll consider those. How about "delegate vars"? |
|
I think I like the most |
|
Same, I think of delegates in C++/C# (i.e., function pointers). |
|
Merging. btw, I tested out this with Compliment and it works there too 🍻 |
Part of clojure-emacs/cider-nrepl#788
Use case explained over there.
...I had attempted this before: clojure-emacs/cider-nrepl#790 but it was targeting the wrong middleware. This impl is simpler / more predictable, needing no
pmap-like parallelism.I've tried this one locally, and it works as intended.
Works on CLJ and CLJS, tests added for both.
Cheers - V