* implementation
* tests
* maybe fix windows compilation?
* undo change of match_flag_type
* switch to boost regex
* move flags to BRegExp object
* fix tests due to cmake 4 update
* update grammar
* update prettifier
* move flag parsing to BRegExp ctor
* add AST nodes, update instr generation and execution
* copy tests but use regular expression literals
* fix grammar for handling division correctly
* modify return values a bit, update tests
- string.match: make groups hold structs of matched, offset
- string.match: add offset
- string.replace: use a groups array like string.match vs individual arguments
* bundle of changes
- standardize error messages
- add more tests for coverage
- move BRegExp creation to static method: previous implementation had
ctor throwing, which was no good inside executor since it didn't have
a try/catch
* more tests; add OT_REGEXP to basic.em
* move string regex stuff to bregexp and support regex/wregex via std::visit
* maybe fix compilation errors?
* add unicode escape sequence handling
* really maybe fix compilation error?
* fix compiler warnings
* refactor a bit to remove duplicate code
* address discord comments
- just return wstring, no need for vector<wchar_t> method
* Add string.split by string and regexp
* Use Max_Split instead of Limit
This makes it match basic::SplitWords
* add support for empty string delim in SplitWords
* remove duplicate code across string.split and mf_SplitWords
* Squashed commit of the following:
commit b3148e069f
Author: turleypol <turley@polserver.com>
Date: Sat Sep 27 17:02:22 2025 +0200
memorylocation of input string is not allowed to changed, switched back
to uniqueptr
commit 5a4d1c8fcb
Author: turleypol <turley@polserver.com>
Date: Sat Sep 27 10:18:01 2025 +0200
no need to use ptr for input
commit 90fa0edc66
Author: turleypol <turley@polserver.com>
Date: Sat Sep 27 09:56:29 2025 +0200
fixed typo
commit bed4ff3ac9
Author: turleypol <turley@polserver.com>
Date: Sat Sep 27 09:47:36 2025 +0200
make Callback for BContinuation move only
* use uninit when group isn't matched
* fix multiline flag handling; add test
* add docs and doc example tests
* some cleanup
* address review comments
- allow move assignment
* add core-changes
* Format escript testsuite
* Manual formatting for escript testsuite
- Add format-off to places where formatter does what it should, but we don't like the results
- Manually change places where formatter outputs weird things _without_ format-off, so next format-all would show changes if not fixed
* Update .err files; add update_err.py helper script
Use the format `:line:col: message` for all .err file checks. This
allows us to change the file names without needing to edit the .err
contents.
Note that `in/inc0{1,2}.err` need have error messages between POSIX and
Windows, so their messages are a bit more "generic" than the other errs.
* Format pol testsuite
* Manual formatting for pol testsuite
* Format pol-core modules
* Manual formatting for pol-core modules
* Additional format pol-core modules
Use FormatterLineWidth=10000 since module functions must be on one line
because of the parse_modules target
* Address Discord comments
- Support multi-line module function declarations in parse_modules
* Format on df07831
The comparison (< and > operators) between objects of different types is
now based on object's addresses in memory. This leads to unpredictable results
and causes troubles with test scripts.
Proposed solution is:
- This is undefined behavior and could change in future, leave it undocumented
- Error and Undefined should always be < than any other object
- Objects of different type should compare < or > based on their Type ID
- Objects of the same type should always have comparison method implemented.
The code should not compile if those methods have not been implemented.
