l...@phdk.org writes: > When adding a newline to an org-src block with > `org-src-tab-acts-natively' set to t, if there's a non-whitespace > character following the pointer at insertion, all other lines are > indented by an additional `org-edit-src-content-indentation' > regardless of their original indentation. > > MRE with -Q on the current dev branch: > Put the pointer between foo and bar in the following src block and > press enter repeatedly. > > #+begin_src fundamental > foobar > #+end_src
Confirmed. The first indent is valid - Org ensures that previously unindented line is indented to `org-edit-src-content-indentation'. However, the following indents are indeed not right. > What's happening is that before indenting, `org-remove-indentation' > removes indentation equal to the least indented line in the block, > which in this case is 0. It then inserts it in the src-edit buffer, > runs the modes indentation, and adds the org-src content indentation > again before reinserting it in the org buffer. > > I've attached the monkeypatch that I cooked up for myself. It passes > tests and hasn't caused any issues for me yet, but Ihor's answers to a > similar sounding report [1] makes me a bit uncertain on what the > intended behavior is. I'm not really sure if might somehow be a > breaking change. If the behavior is unintended but this patch is > breaking, an alternative could be making `org-src-tab-acts-natively' > always reindent the entire block, but that seems overkill to me. Note that in the previous report I was only talking about the first indent. I did not anticipate that repeated <RET> will cause such breakage. > ;; rather than `indent-line-function' here or we may > ;; sometimes break `electric-indent-mode' > ;; > https://orgmode.org/list/5O9VMGb6WRaqeHR5_NXTb832Z2Lek_5L40YPDA52-S3kPwGYJspI8kLWaGtuq3DXyhtHpj1J7jTIXb39RX9BtCa2ecrWHjijZqI8QAD742U=@proton.me > + (save-excursion > + (beginning-of-line) > + (insert-char #x20 block-content-ind)) This looks about right. May you please: 1. Use `insert' rather than `insert-char' that is for interactive use. 2. Use more readable ?\t notation 3. Add a comment linking to this report and explaining why we need it 4. Write a simple test so that we avoid future regressions -- Ihor Radchenko // yantar92, Org mode maintainer, Learn more about Org mode at <https://orgmode.org/>. Support Org development at <https://liberapay.com/org-mode>, or support my work at <https://liberapay.com/yantar92>
