sets the stroke-width to 3. This leads to weird rendering in edge cases (e.g. using print-to-PDF the output will look extremely bold in most PDF readers even if it will print fine).
I recall we discussed this for v2 ages ago and I think the original reason was to cover for flaky SVG rendering in old browser engines.
We've been using 0 in production for years now (in v2) and I just noticed that I had missed this declaration when we switched to v3 last year.
Running a small random sample through visual regression testing, it looks like switching to 0 can affect some stretchy constructions (for the better - they come out a hair thinner especially which improves where they connect with non-stretchy pieces). I didn't test ancient or dead browser engines but I used to test down to IE9 so I'm relatively certain this won't cause significant problems.
https://github.com/mathjax/MathJax-src/blob/2dd53ce6c8af3c9cceba0baf014ea9b065130774/ts/output/svg/Wrappers/TextNode.ts#L49-L53
sets the stroke-width to 3. This leads to weird rendering in edge cases (e.g. using print-to-PDF the output will look extremely bold in most PDF readers even if it will print fine).
I recall we discussed this for v2 ages ago and I think the original reason was to cover for flaky SVG rendering in old browser engines.
We've been using 0 in production for years now (in v2) and I just noticed that I had missed this declaration when we switched to v3 last year.
Running a small random sample through visual regression testing, it looks like switching to 0 can affect some stretchy constructions (for the better - they come out a hair thinner especially which improves where they connect with non-stretchy pieces). I didn't test ancient or dead browser engines but I used to test down to IE9 so I'm relatively certain this won't cause significant problems.