Writer HTML import messing up font size (regression 7.6)

Emerging from this post, seems Writer (7.6.5.2) opening a HTML with
<font face="Times New Roman, serif"><font size="7" style="font-size: 36pt">Text 36</font></font>
assigns it a 12pt size :frowning:

(the HTML example was actually produced by Writer itselft, “Save as”)

any idea of a recent change causing that ?

Had a look at your other post in French. Can’t figure out your goal.

Are you using Writer to write HTML code? This is not the way to do it. You write a document with usual formatting directives (normally styles) and this formatted document can be exported to HTML where styling information is translated to CSS.

If you write directly HTML code, you’ll be better off with a text editor.

For better understanding of the issue, describe your workflow (initial data: where and how is it stored? which processing on this initial data? …).

  • start Writer
  • New document
  • Type … “your text
  • Select “your text
  • Change size to 36pt (or whatever)
  • Save As HMTL “yourFile.html
  • Close
  • Open “yourFile.html

https://bibisect.libreoffice.org/

@fpy: using your procedure, everything behaves as expected up to Save As. When I open the file in Firefox, it displays as specified.

When I reopen it in Writer, any direct formatting is ignored. All formatting provided by styles, even user styles, is preserved.

This may be an insufficiency or bug in the import filter but, as always with Writer, direct formatting is a wide door to formatting hell. Learn to work with styles. You’ll gain comfort, versatility and efficiency.

was it documented ? where, since when, … ?

thanks. I may even teach.
HOWEVER … we’re talking about regression here … which may well be a far worse hellish door :wink:

@mikekaganski 94ff994d396ae38edf4f72215452b4c4b086490d is the first bad commit

not in core ?

Did you read the bibisect documentation, which mentions, that the core commit is mentioned in the bisect commit message?

nope. me not a dev here.

Then I suggest to copy and paste the whole output of the bisect result. (I think that the bibisect documentation is written to target general audience, explaining things to non-developers, but shrug…)

(hum. thought I tried this one too … maybe copy/paste error)

ok so, blame here :
https://git.libreoffice.org/core/+/db115bec9254417ef7a3faf687478fe5424ab378

reopen ?

No. Create a new bug; describe the problem; write that it is bibisected to that commit; and add that bug to See Also.

piggybacking on tdf#157411.