Open ronaldtse opened 1 year ago
In the content tree there is text for x
and y
:
Copy-pasted text also contains them: arguments y and x, which
.
Currently, mn2pdf inserts hidden math as the transparency text (https://github.com/metanorma/metanorma-bipm/issues/188). Looks like, Acrobat doesn't see such text in the comparison feature.
I'll try to remove transparency mode temporarily locally and test the comparing result.
Thank you for investigating! Indeed this is strange. I can verify in Preview that I can copy and paste this text.
I've tested with different combination of color + transparency for hidden text:
FFFFFF
+ transparency mode (currently) - there are differences in the Acrobat comparison results (i.e. Acrobat doesn't see the hidden text)FFFFFF
without transparency - there are differences in the Acrobat comparison results (i.e. Acrobat doesn't see the hidden text)FEFEFE
(almost 'white') without transparency - no differences in the Acrobat comparison results for y and x
FEFEFE
(almost 'white') + transparency mode - there are differences in the Acrobat comparison results (i.e. Acrobat doesn't see the hidden text)I.e. if the text in white color or/and transparent, then Acrobat ignore it in the compare feature. But it's available for copy-paste feature. I don't figure out which is workaround solution can be applied...
Then let's just keep it as it for now. I don't think we do comparisons too often...
This is an interesting topic that @stuartgalt would be interested in as the PDF guru...
Similar to #870 this is a problem with Adobe Acrobat's Compare PDF feature. Letting @stuartgalt know in case the PDF TC has (or plan to have) specs for Compare PDF.
In PDF diffs it is important that the content does not introduce systematic changes from the original PDFs.
In the Metanorma-generated PDFs, the math content is missing from the content text. This is a sample from ISO 10303-50: (https://github.com/metanorma/iso-10303-detached-docs/tree/main/sources/iso-10303-50)
In the generated PDFs, all the formula contents are "missing" from the content text. Given that we can insert AsciiMath for it, we would reduce a lot of these false positives.