I often use Splice & Kill/Delete Backward where Raise Sexp would be more appropriate. Good tip! I'll try to remember to use that instead of ctrl+alt+shift+backspace in future!

Thanks for sharing, both of you! 🙏 It's funny with raise sexpr. When I learnt about it, I wondered why on earth I would ever want to do that? Then I find myself raising sexprs all day. 😃 I also use splice and kill backwards a lot. Two power edits, for sure.

I even use convolute (occasionally!)...

Respect!

Raise Sexp. Demo in the post right before yours. Replaces surrounding form with either a selection or the next form after the cursor. https://clojurians.slack.com/archives/CBE668G4R/p1668129301017559

As in... don't barf first, just use Raise Sexp for the entire action.

But, in case you aren’t keeping anything (meaning, Raise Sexp wouldn’t be used), I have the following to Delete Enclosing Sexp: • installed macro-commander extension • macro config:

"macros": {
    "paredit.killCurrentSexp": [
      "paredit.backwardUpSexp",
      "paredit.killSexpForward"
    ],

As for your case of having already barfed, Splice & Kill/Delete Backward works to clean up the leftover form.

@U90R0EPHA Splice & Kill/Delete Backward This is what I wanted. I missed it somehow :S

Note that with Splice Kill Back, any forms to the right of the cursor, but inside of the current one (the one being killed) are spliced up instead of being deleted with the form. (And Splice Kill Forward does the opposite.

((:foo |:bar))  --> (|:bar)  ;; SKBack
((:foo |:bar))  --> (:foo |) ;; SKForward

@U90R0EPHA Actually I'd like a command that'd delete the entire enclosing sexp

I think you need 2 commands as @U08TWB99B demonstrates in his macro above. I think the commands he chose will give inconsistent results, but either of these combos should work consistently: Kill/Delete Forward to End of List -> Splice & Kill/Delete Backward or Kill/Delete Backward to Start of List Back -> Splice & Kill/Delete Forward Kills anything that would be spliced, and then kills the rest, splicing nothing into the surrounding form. The List-Forward-> Splice-Back variation feels pretty natural to me, and I will probably start practicing it. You can pretend it is a key chord instead of two separate commands.

What I have been doing is Move Back Up Sexp -> Select Forward Sexp -> Backspace. But this seems more direct.

I guess you put a tap> or prn here. Unless I'm mistaken.

(let [stats (merge default rest?)
        body (->> stats (m/encode "application/edn") (tap>) (HttpRequest$BodyPublishers/ofByteArray))])

I would think there'd be a Calva shortcut for this @UC1DTFY1G, no? Usually when I'm working with the various thread macros I can use Ctrl+Alt+Enter but that does not seem to play nicely with let

Though I guess stats isn't defined until that closing ]... is that where the issue lies?

@U03QBKTVA0N I guess that whatever ctrl+alt+enter is doing doesn't know about stats.

@U03QBKTVA0N Sorry didn't read your last comment. Yes, I guess so.

(let [stats (merge default rest?)
        body (->> (range 10) (map inc) | (HttpRequest$BodyPublishers/ofByteArray))])

I figured as much

I do wonder if there's a way to do this though

Unless I misunderstand something, this is where Evaluate Selection, Closing Bracket comes in. See https://calva.io/evaluation/#evaluate-selection-closing-brackets

No, evidently this is what I need. I was only selecting up until [ and not back to (let ... thank you @U0ETXRFEW!

Funny, it's not working for me

I think it would work but it doesn't know what default is then, right @UC1DTFY1G?

(let [r (range 10) 
      s (->> r (map inc) | (filter odd?))]
  s)

ugh

Which version of Calva are you using, @UC1DTFY1G?

Ok maybe I'm mistaken too, I get nil instead of a syntax error :thinking_face:

@U0ETXRFEW [2.0.293] - 2022-08-18

Well, the let will not be returning anything there. nil is to be expected.

@UC1DTFY1G, is ”does nothing for me” the same as ”returns nil”?

No, doesn't even return nil. I am updating. let's see

This is what you should get:

Because what is evaluated here is:

(let [r (range 10)
      s (->> r (map inc)])

Yeah, that makes sense @U0ETXRFEW. I guess the easy workaround here which I just did is to use #_ on that last form like so

(let [stats (merge {:a 1} {:b 2})
        body (->> stats (m/encode "application/edn") #_(HttpRequest$BodyPublishers/ofByteArray))]
    body)

Yeah, returns nil after update

and then I can see what I want to see

2.0.293 is ancient! 😃

[2.0.308] - 2022-10-19 • https://github.com/BetterThanTomorrow/calva/issues/1901 • https://github.com/BetterThanTomorrow/calva/issues/424

That was the version I had installed in my WSL2. Somehow the WSL2 version didn't auto update.

We should consider a command that did what you tried to do with that let binding, @UC1DTFY1G. I wonder what it would be like...

Re-write and eval like this?

(let [r (range 10)]
  (->> r (map inc))

I should look at this 😄

Let me see how other commands are implemented in Calva and perhaps I can figure out something...

I guess your original suggestion about prn would be useful here as well, though I have not yet tried it

I think tap> is more useful perhaps

Especially if you are using something like portal

Yeah, just got into Portal the other day. It's great

Yeah, it's really nice. Though I don't much like the single click zoom-in

Maybe some way to insert tap> before closing brackets?

ok, that'd work

I’m late here, but one thing I might do in a case like this is use an inline def with the same name as the binding.

(let [stats (merge default rest?)
      _ (def stats stats)
      body (->> stats (m/encode "application/edn")(HttpRequest$BodyPublishers/ofByteArray))])

It’s not something you want to leave in your code (clj-kondo will help you remember not to), but it’s very helpful for evaluating smaller pieces of code in functions.

I would never have thought to do this but that is quite useful. Thanks @U9A1RLFNV

Yeah, I learned this a while back and I use it often now.

And, you’re welcome

Inline def is a power move. I use it all the time. It will litter down your namespace, which seldom is a problem, but could be good to have ns-unmap ready for use. 😃

@U03QBKTVA0N maybe this video, that just aired, could interest you. It shows me using inline def to inspect the application. https://www.youtube.com/watch?v=fMtvtAfdr90

If I put member on the next line, it's even weirder:

new-user (-> (field/as-user db-spec new-user-id)
                               (as->
                                member
                                (settings/update! (:database app)
                                                  member
                                                  :view "s")
                                 (settings/update! (:database app)
                                                   member
                                                   :like "s")))

member and the first clause get one space, the second (and any subsequent) clause(s) get two spaces.

I've tried enabling/disabling calva-fmt and I've tried the New Indent Engine and they all seem to give the same layout.

That first one looks like it only went through the more lax "as you type" formatting. I would expect pressing Tab to line up the second settings/update! with the first. Is that not the case? Other than that, looks to me like what I normallly see, I think.

It is not the case. I have pressed tab on every part of that form, trying to see if it will reindent things.

I think the formatting rule for as-> is just broken.

Actually all that stuff lining up perfectly with (as-> is strange, should be some indent there

Oh, there is a 1 char indent ?

I suspect whoever implemented the as-> rule looked at as-> expr sym clauses* and based it off that instead of thinking about how as-> is supposed to be used (inside ->, so that first expr will be missing).

Re: "1 char indent?" -- as I said above: member and the first clause get one space, the second (and any subsequent) clause(s) get two spaces.

FWIW, cond-> also gets bad indentation inside ->:

new-user (-> (field/as-user db-spec new-user-id)
                               (cond->
                                (settings/update! (:database app)
                                                  member
                                                  :view "s")
                                 (settings/update! (:database app)
                                                   member
                                                   :like "s")))

Again, 1 char indent for the first clause, 2 char indent for all the rest.

I think this is expected from cljfmt. as-> has the same default setting of [[:block 2]] as condp, which treats the first 2 arguments as special.

(condp :foo :bar
  :baz
  :buz)

(condp :foo
       :bar
  :baz
  :buz)

(condp
 :foo
 :bar
  :baz
  :buz)

So it's basically wrong 🙂

Both as-> and cond-> (and probably a few others) should have different indent behavior inside -> pipelines.

I don't think there is a way to express as-> inside -> in the indentation rule format for cljfmt... I could be wrong, I am certainly not an expert on the subject.