kakoune.git - Unnamed repository; edit this file 'description' to name the repository.

Age	Commit message (Collapse)	Author
2025-04-02	Tweak ranked match behaviour to consider the number of full words	Maxime Coste
	Tracking the number of query words that appear as full words in the candidate seems to fix a few cases where the existing fuzzy matching algorithm was not great. I have been running with this for a while and did not notice any annoyances, the whole RankedMatch code probably deserves more attention but this seems to go in the right direction.
2023-10-25	Speed up WordSplitter	Maxime Coste
	Only do utf8 decoding once per codepoint instead of twice, limit the byte length instead of the codepoint length.
2020-11-07	Fix performance issue with word completion	Maxime Coste
	When pasting many words with <a-p> we can end-up with a huge concatenated word and many selections, the previous code ended up iterating from each selection cursor to that word start and end to find the word under the cursor. This could lead to performance issue as each selection would trigger iteration on that huge word. This is unnecessary as word completion has a word length limit, so we now take it into account to avoid iterating to far from the cursor position.
2019-08-31	Limit the amounts of calls to get_extra_word_chars	Maxime Coste
	This is a relatively expensive function as it needs to resolve an option, which uses dynamic_cast.
2019-01-24	Header and dependency cleanup	Maxime Coste

2018-04-19	Move get_word_db to word_db.cc	Maxime Coste

2017-05-22	Remove virtual destructor from OptionManagerWatcher	Maxime Coste
	We should never destruct anything through an OptionManagerWatcher pointer, so having all those destructor virtual makes no sense.
2017-03-16	Fixes some clang-tidy warning and add a few missing meta.hh include	Maxime Coste

2017-03-07	Replace uses of UnorderedMap with HashMap	Maxime Coste

2017-01-08	Apply clang-tidy modernize to the codebase	Maxime Coste

2016-08-24	Introduce the 'completion_extra_word_chars' option	Maxime Coste
	This string option is used to get all the additional characters that should be considered as "word" character for the purpose of insert mode completion. Fixes #758
2016-03-25	Move UsedLetters with RankedMatch	Maxime Coste

2016-02-05	Get rid of SharedString	Maxime Coste

2016-01-28	Small cleanup	Maxime Coste

2015-10-22	Extract WordDB::RankedWord as RankedMatch in its own file	Maxime Coste

2015-10-21	Remove legacy WordDB::find_matching method	Maxime Coste

2015-10-18	Experiment with ranked word completion depending on word boundaries	Maxime Coste

2015-03-10	Tweak WordDB implementation	Maxime Coste

2015-03-01	Use StringDataPtr alias for RefPtr<StringData>	Maxime Coste

2015-03-01	rename StringStorage to StringData	Maxime Coste

2015-02-19	Rename safe_ptr and ref_ptr to SafePtr and RefPtr	Maxime Coste

2015-01-27	Store direct ref_ptr<StringStorage> for WordDB lines	Maxime Coste
	Storing a SharedString is a waste, we want the whole line.
2015-01-15	Do not store lines to word in word db, just keep the old lines alive	Maxime Coste

2015-01-15	Replace InternedStrings with SharedString, shared_ptr based	Maxime Coste

2015-01-13	Style tweak in WordDB	Maxime Coste

2015-01-12	replace all std::vector with Vector	Maxime Coste

2015-01-11	Track some more memory	Maxime Coste

2015-01-07	Add initial memory domain allocation tracking support	Maxime Coste

2014-12-23	Remove trailing blank lines	Maxime Coste

2014-12-23	Code refactoring in WordDB	Maxime Coste

2014-12-22	Minor cleanup	Maxime Coste

2014-12-16	Rework hashing, use a more extensible framework similar to n3876 proposal	Maxime Coste
	std::hash specialization is a pain to work with, stop using that, and just specialize a 'size_t hash_value(const T&)' free function.
2014-10-28	Refactor word_db, use an unordered map	Maxime Coste

2014-10-01	Add support for interned strings	Maxime Coste
	Use interned strings for Modification contents and word database. Interned strings are guaranteed not to move in memory and are reference counted.
2014-07-30	Insert word completion: Use subsequence matching if prefix failed	Maxime Coste

2014-05-14	replace LineChangeWatcher with a free function compute_line_modifications	Maxime Coste

2014-04-22	Preserve current word in word completion if found elsewhere	Maxime Coste
	If occurence count in the buffer if greater that one, do not remove it from the matches.
2014-01-27	WordDB now uses a LineChangeWatcher based implementation	Maxime Coste

2014-01-19	WordDB: use an ordered map for storing words	Maxime Coste
	This way we can use lower_bound to find the first prefix match in logarithm time and we know all other prefix matches will follow.
2014-01-16	Word completion use a WordDB stored as a buffer value.	Maxime Coste