Releases: toxicphreAK/docxtpl-ng
Releases · toxicphreAK/docxtpl-ng
Release list
v0.2.0: release: prepare 0.2.0
Fifteen issues (#28–#42), filed after reading the whole upstream docxtpl tracker and asking which of its open reports this architecture could answer. - `__version__` becomes `0.2.0`, and `HISTORY.md` gains the heading and date. A minor bump rather than a patch: twelve of the fifteen add public API. The release notes lead with the three that turned out to be worth more than the report that prompted them — the render cache, which came from measuring that Jinja2 compilation is 86% of a render where the phases everyone assumes are the problem are 5%; the run-style check, which had to move from where the issue proposed it because a Text holds no template; and lists, where defining numbering per list dissolved both constraints that looked unavoidable. They also state the one behaviour change plainly rather than burying it: `StyleError` for a run style the document cannot carry. Only styles a render *introduces* are checked, so nothing that renders today starts failing — but a template that was silently producing unstyled text now says so, and someone upgrading should know that before they do. Verified: 479 tests pass on 3.12, ruff and pyright are clean, mkdocs builds in strict mode, and `make check` reports both distributions well formed. Pushing this does not publish anything — that takes a `v0.2.0` tag. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
v0.1.1: docxtpl-ng 0.1.1
Two bugs a loop caused — duplicate drawing ids, and a w:tblGrid that did not
follow a {%tc %} loop — and the rest of the docxtpl public surface.
See HISTORY.md.
v0.1.0: docxtpl-ng 0.1.0
First release. Renders .docx files as Jinja2 templates: the full docxtpl tag dialect, every part that can carry a tag, the content objects, and a compatibility surface that makes swapping the dependency the whole migration. See HISTORY.md for what is in it.