Neil Hodgson

nested square brackets, TeX files

Issue #360 is for the latex lexer and this bug is for the tex lexer.

Committed [fa2ab2] change that handles null pTextFormat and clamps weight and stretch. There are different places to perform the clamp with a choice between safety, precision, and allowing some edge cases that may cause reasonable behaviour in some situations. There is a documented stretch value DWRITE_FONT_STRETCH_UNDEFINED = 0 that always caused E_INVALIDARG for me on Windows 10. It may work somewhere, perhaps as a wildcard that matches any available stretch. Only clamping inside the retry replaces...

The standard technique to cancel a drag is to press the escape key. While some programs behave in this way, it isn't universal. It isn't, for example, implemented in Qt Creator or Chrome. The Visual Studio implementation invokes the context menu so may be an accident of the context menu implementation.

While a clarifying note may be helpful, mentioning EnumFontFamiliesEx anchors the explanation in legacy GDI terms and concepts. In general, DirectWrite should be treated as the default and GDI as a legacy API with some differences. Eventually GDI will be depreacted.

I am not currently accepting LLM-generated contributions. The presentation of this feature, with 100 new setting combinations, is overwhelming so will not be that helpful for users.

The portion of DrawEOL is clipping rcSegment to rcLine and it may be better to make that explicit with a call to Intersection or similar. I considered making a change in this direction for [dbd858] but was uncertain about best formulation and it was close to next release. The DrawIndentGuidesOverEmpty changes are, to me, adding clutter without sufficient benefit. The small reduction of work will be overwhelmed by other actions performed.

Add API `SCI_FINDCOLUMN(line line, position column, int tabWidth) → position`