Date: Fri, 14 Jun 2024 21:17:22 +0300
Cc:steven@stebalien.com,jporterbugs@gmail.com,71525@debbugs.gnu.org
From: Dmitry Gutov<dmitry@gutov.dev>
On 14/06/2024 10:13, Eli Zaretskii wrote:
/* Insert a sequence of NCHARS chars which occupy NBYTES bytes
starting at GAP_END_ADDR - NBYTES (if text_at_gap_tail) and at
- GPT_ADDR (if not text_at_gap_tail). */
+ GPT_ADDR (if not text_at_gap_tail).
+
+ If BEFORE_MARKERS is true, insert before markers. */
This commentary should also mention process.c as the single caller
using this facility in unconventional ways.
Also note that I did make some effort to make the field
coding->insert_before_markers used consistently: in the changeset, it's
not just obeyed on the code path that decode_coding_c_string goes
through - it's used anywhere where insert_from_gap is called and a
coding_system structure is available - e.g. in encode_coding.
Also in decode_coding_gap, but that one seems to only have one caller
(insert-file-contents), so that choice is moot.
Something else I've noticed (a bit off topic): insert_from_gap_1's
comments says that it "does not invalidate any cache, nor update any
markers, nor record any buffer modification information of any sort",
but then it calls treesit_record_change when available. Not sure we need
to change that, but maybe we should update the comment, at least.