stray `:END:` markers are duplicated on `headline-set-node-property`

[hrehfeld]

Edit: again, this seems to be an org-element bug? The org-format is also not very strict and/or explicit on this...

Evalute

(org-ml-update* (org-ml-headline-set-node-property "TEST" "TEST" it) (org-ml-parse-this-subtree))

on

* test

:END:

results in:

* test

:PROPERTIES:
:TEST:     TEST
:END:
:END:
:END:

Subsequent evaluations of the same quickly spiral into oblivion:

* test

:PROPERTIES:
:TEST:     TEST
:END:
:PROPERTIES:
:TEST:     TEST
:END:
:END:
:END:
:END:
:END:

$ ls -alh
-rwxr-xr-x 1 hrehfeld hrehfeld 4.1M Oct 31 22:59 test.org

(org-ml-parse-this-subtree) and (org-element-parse-buffer) parse a drawer named END:

(headline 
  (:raw-value "TEST" :begin 1 :end 15 :pre-blank 1 :contents-begin 9 :contents-end 15 :level 1 :priority nil :tags nil :todo-keyword nil :todo-type nil :post-blank 0 :footnote-section-p nil :archivedp nil :commentedp nil :post-affiliated 1 :title (#("TEST" 0 4 (:parent #0))) :parent nil) 
  (section (:begin 9 :end 15 :contents-begin 9 :contents-end 15 :post-blank 0 :post-affiliated 9 :parent #0) 
  (drawer (:begin 9 :end 15 :drawer-name "END" :contents-begin nil :contents-end nil :post-blank 0 :post-affiliated 9 :parent #1))))

Interestingly, this doesn't duplicate:

(org-ml-update* #'identity (org-ml-parse-this-subtree))

but this does:

(org-ml-insert-tail (point) (org-ml-parse-this-subtree))

So does this:

(org-ml-update* (org-ml-headline-set-title! "TEST" nil it) (org-ml-parse-this-subtree))

Maybe because the first doesn't update at all?

[ndwarshuis]

This seems like an upstream error. It doesn't make sense for org-element to parse ":END:" (without another ":END:") as a drawer named "END". The reason why the identity doesn't do it is because there is a check in org-ml-update that will noop if the final output is the same as the input.