Thank you for sharing your kind and funny jokes.
Let me remind you that in a free world there exists something called the freedom of choice. I did not write these two pieces of software, but I think their authors who provide freely their work deserve better than some absurd sniggers. I even wish some of you sincerely to be as computer savvy as them.
So what?
Linux Libertine.
I chose it for two reasons: first because it's my reading font of choice (I agree here with doubleshuffle and I doubt we may be alone). Second because I wished to provide an example and needed for that to use a font family with a permissive license (here GPL) which allows to rename and modify these fonts.
The subset tool.
If you have complex technical questions about it, I am sorry to be unable to answer them because I am not one of the authors but feel free to contact them. You nevertheless can test the EPUB text display on your ebook readers. I shall also point some facts and differences that may interest some users:
- I found that the Calibre subset tool could not work with the Chinese ttf font STSong (15,4 MB). pyfysubset did it in a split second. I cannot attach this Chinese font here because I do not know its license.
- I found when dealing with Linux Libertine that the Calibre subset tool produces on average subsets which are about
three (ttf fonts) or
four (otf fonts) times bigger than the ones produced with pyftsubset. Feel free to make your own experiments.
Not a real subset
I have been told that:
Quote:
|
The problem here is that Calibre's editor doesn't differentiate between regular, bold, italic, & bolditalic. So you don't get proper subsetting. You get get such by using the Calibre to do the subsetting.
|
I am afraid you are wrong on this one. It happens that I already asked precisely Kovid Goyal if he used for his subset tool a tailor made list of italic, bold, etc. characters (
https://www.mobileread.com/forums/sh...d.php?t=291401). He does not. So, the same global list of characters is also processed several times...
From there, it does not mean that pyftsubset is a superior technique. Just that it deserves to be further tested, and it's only what I propose to do.
A slow technique
For the time being, the technique used for producing this subset is obviously slower than the one used with Calibre. I hope you are aware that this thread belongs to a workshop and that the current method is not to be used for ever.
For example, if somebody from the Sigil team shows later some interest in it, it's will be an easy task to make a plugin out of it or even to integrate it exactly the way it is with the Calibre editor. The list of characters is already produced both by the Calibre editor and Sigil. After that you just need to launch a single command for each font... It's really a basic stuff.
Please, forgive me if I spoiled the party. Follow on please, some of you were joking about square wheels...
