[okular] [Bug 506426] WISH: Avoid full font-embedding with commenting tools

[Sune Vuorela]

https://bugs.kde.org/show_bug.cgi?id=506426

--- Comment #6 from Sune Vuorela <kde at pusling.com> ---
(In reply to Klaus from comment #5)
> ________________________________________
> WORKAROUND
> 
> It is currently possible to edit “okularpartrc” located e.g. in
> 
>     ~/.var/app/org.kde.okular/config/  # for flatpak
>     ~/.config/okularpartrc    # for regular installations
> 
> to achieve the wanted behavior.
> 
> ________________________________________
> STEP BY STEP
> 
> 1. In Okular, change the font for new typewritter and inline notes to
> anything but the default. Also adjust the font size, as changing it later
> will require editing the “okularpartrc” file again. E.g. “Monospace” for
> typewriter comments and “Sans Serif” for inline note comments.
> 
> 2. Open the “okularpartrc” file and replace the font names chosen above by
> one of “Helvetica”, “Courier”. I didn't test whether you need “Times Roman”
> or “Times” for a serif font.
> 
> Now inserting new comments uses standard PDF fonts.
> 
> ________________________________________
> LIMITATIONS
> 
> 1. When you reopen or reload the file later and edit the text content of the
> comment, that comment is switched to the default GUI font, i.e. this
> workaround is largely sabotaged by an unrelated bug. On Windows 11, the
> default font is "Segoe UI Emoji", which is HUGE (several MB). On my Linux
> system it is “Noto Sans”, which still adds 300 KB by itself. Since the fonts
> are fully embedded and not as subset fonts, you can quickly end up with very
> large PDFs when commenting. To avoid this you have to avoid editing existing
> comments, and instead have to rewrite them (augmented by cut+paste). Within
> the same session, editing works without that issue.
> 
> 2. Font embedding does not work consistently. Sometimes the font gets
> embedded despite having a standard PDF font name, sometimes not. Might be
> related to font substitution.

Please note that this also requires newest Poppler (25.10)

And yes, if we ask to embed "Helvetica" and we get something actually
"Helvetica", we still embed it (for all the 14 base fonts). We only skip the
embedding in poppler if we get a substitute font.

I do think that font-subsetting (and something slightly smarter font
refcounting in poppler) is the real fix here, but please don't hold your breath
for that.

-- 
You are receiving this mail because:
You are the assignee for the bug.