Re: Exposing buffer text modifications to Lisp

[Eli Zaretskii]

> From: "Basil L. Contovounesios" <contovob@tcd.ie>
> Cc: Ihor Radchenko <yantar92@gmail.com>,  casouri@gmail.com,
>   emacs-devel@gnu.org
> Date: Wed, 22 Jun 2022 18:45:53 +0300
> 
> Would any of Gnulib's generic container types[0] be appropriate in this
> case, and if so, fair game for using in Emacs' sources?
> 
> [0]: (info "(gnulib) Container data types")
> https://gnu.org/s/gnulib/manual/html_node/Container-data-types.html

Could be.  But I think we need more research before we decide.
Someone™ should study the usage patterns of the markers for
character-to-byte translation, and see which operations should be the
fastest and which could be slower.  Armed with that information, we
could then select the best data structure for the job.

Btw, do we have recipes for measuring the effects of changing the data
structures used for markers?  If we do have such recipes, did someone
try to compare the performance in plain-ASCII Org buffers (where the
conversion is trivial and shouldn't even access the markers) and
non-ASCII buffers?  Inserting a single non-ASCII character somewhere
in an otherwise plain-ASCII buffer should show the effect of many
markers on the likes of CHAR_TO_BYTE.