As discussed on https://github.com/mathjax/MathJax/pull/1723 a follow up change to
T105310
could be to use improved sizing operation as explained here
https://github.com/mathjax/MathJax/issues/566#issuecomment-257555181
.
I think that's something nice to have even though it would take a while until all images are re-generated.
Any objections?
Sounds like an interesting approach to take. @Physikerwelt have you tested the proposed solution? If so, I'm game to move on this.
@mobrovac not yet, but I'll get to it. I'm currently testing the latest MathJax-node version. While in theory everything works smoothly, I see different result in the test.
Expected: "vertical-align: -0.338ex; width:9.025ex; height:2.676ex;" Result: "vertical-align: -0.338ex; width:8.976ex; height:2.676ex;"
Maybe it's best to update the functions to the new API and care about the tests in a second step.
Is just the bounding box different or has the scale changed? I agree it's best to update the functions first, but let's make sure there are no regressions.
Updating all the test to the latest mathjax node version took a while. But it's done now. https://github.com/physikerwelt/mathoid-server/tree/sre-testing I'll wait with the pull request though until I did run the regression and performance tests