The "interpolated snippet" in LuaSnip.

[no-vici]

Hi LuaSnip enthusiasts,

I had these two snippets a while ago, the former works as expected for its simplicity, and for some arbitrary reasons I moved to the latter one and struggled to interpret its meaning. I asked ChatGPT and I got this answer.

1. s("do", fmt("do\n\t{1}\nend", i(1, "--do_body")))
2. s("do", { isn(1, t({ "do", "" }), i(1, "--do_body"), "") })

The logics from the AI's answer sound reasonable, the 2nd snippet worked before but now I got the error bellow. Could someone kindly explain to me what are the advantages of the second snippet over the first one and what the error means please?

[L3MON4D3]

Whew, maybe our jobs are safe for a while yet :P
isn is an indentSnippetNode, it works like a snippetNode, but accepts an additional parameter which dictates the indentation of the nodes (look for it in DOC.md for more details) (but it seems unnecessary in this case.?)

In any case, it should work if the nodes passed to the isn are wrapped in a list

[no-vici]

isn is an indentSnippetNode.

I would blame the AI for this ;) I've had a looked at the DOC.md for a while, forgotten most of the concepts.

In any case, it should work if the nodes passed to the isn are wrapped in a list.

The syntax looks right, right? I will come back to the original version as it supports indented blocks nicely, it did not before, and that was why I looked up for the second version.

Thank you for looking into this.

Cheers.

[L3MON4D3]

You're welcome :)

Quickly, why 2 has no indenttion: the thing with isn is that all indentation, even the indentation of the trigger is disregarded (this is so that text can be inserted at column 0, regardless of the triggers indentation (probably useful for stuff like #ifdef in c/c++)), and it can be reinstated by passing "$PARENT_INDENT" to the isn.
But version 1 looks perfectly fine, version two would only be more elegant if there were additional lines in the do-end-block

[no-vici]

... and it can be reinstated by passing "$PARENT_INDENT" to the isn.

Oh, I see, didn't think about the "$PARENT_INDENT". I will try this.

Thank you once more.

[no-vici]

... it can be reinstated by passing "$PARENT_INDENT" to the isn.

Wow, that does the trick. Thank you, thank you.

s("do", { isn(1, t({ "$PARENT_INDENT" .. "do", "$PARENT_INDENT" }), i(1, "$PARENT_INDENT" .. "--do_body"), "$PARENT_INDENT" .. "" ) })

[L3MON4D3]

Anytime :)