| FazBrowse GitHub Viewer | Trending | | Home |
| Tools: [Download Repo ZIP] [Original HTTPS Page] |
Sorry, something went wrong.
|
Off topic: should I file a bug report for dvipng? (The author, coincidentally, is a colleague, currently my boss...) |
Sorry, something went wrong.
|
I don't think it's really a bug in dvipng, just a different definition of "baseline". OTOH one thing that we would look into is to use dvipng --depth... to get the baseline position? But that would require going through dvipng even for the vector backends. |
Sorry, something went wrong.
Should we return to the plan of reverting the original PR? |
Sorry, something went wrong.
|
Let's do that for now. |
Sorry, something went wrong.
|
The original PR was reverted, so I assume you'll just put this into 3.7. |
Sorry, something went wrong.
|
Yes. |
Sorry, something went wrong.
|
Moved to 3.8, but feel free to argue for a back port. |
Sorry, something went wrong.
|
No hurries, this still needs some work... |
Sorry, something went wrong.
| Back | FazBrowse Home | New Git URL |
PR Summary
Closes #23977. One can also check with the simpler example given at #23977 (comment).
Actually in the specific case of the eqnarray given in #23977 there would still seem to be an extra blank line at the bottom, but this is due to the extra \\ at the end of the eqnarray, which is now taken into account (... but not by dvipng apparently (*)); removing it (as shown in documented eqnarray examples, e.g. http://latexref.xyz/eqnarray.html, https://texfaq.org/FAQ-eqnarray) handles that.
(*) but hopefully we'll get rid of dvipng anyways in favor of directly rendering the glyphs ourselves; see e.g. #22459 (comment).
Edit: no, that's not enough if the last line has nonzero descent. Back to the drawing board.
PR Checklist
Tests and Styling
Documentation