POP not follow the Atomic rule

[huacnlee]

mark = ${ PUSH(mark_wrap) ~ mark_string ~ POP }
mark_wrap = { "***" | "**" | "*" | "~~" | "__" | "`" }
mark_string = {(!(PEEK) ~ ANY)* }

With text:

**H.263**

Output

- mark
  - mark_wrap: "**"
  - mark_string: "H.263"

Expected

- mark
  - mark_wrap: "**"
  - mark_string: "H.263"
  - mark_wrap: "**"

[tomtau]

@huacnlee not sure if this is a bug or if this is fixable without breaking changes. POP is a builtin rule which behaves in that way (similarly to PEEK or if one e.g. had ANY* inside an atomic rule).
One potential workaround is to introduce another rule, e.g.:

mark = ${ PUSH(mark_wrap) ~ mark_string ~ mark_wrap_pop }
mark_wrap_pop = { POP }
...

[huacnlee]

Ok, I got it.

[huacnlee]

I have met this problem again.

attr_quoted =  { PUSH(quote) ~ attr_value ~ POP }
attr_value  =  { (!PEEK ~ ANY)* }
quote       =  { "\"" | "'" }

Get

- attr_quoted
  - quote: "\""
  - attr_value: "https://google.com"

Why the PUSH can return value, but POP not?

PUSH and POP is a pair, why not them do the same behavior?

And if we write this with:

attr_quoted =  { PUSH(quote) ~ attr_value ~ end_quote }
end_quote = { POP }
attr_value  =  { (!PEEK ~ ANY)* }
quote       =  { "\"" | "'" }

The pest syntax is looks like not too good.

@tomtau

[tomtau]

Why the PUSH can return value, but POP not?

PUSH and POP is a pair, why not them do the same behavior?

These decisions were a long time ago; not sure if @dragostis, @CAD97 or anyone from @pest-parser/triage can recall it?