CI workflow to generate the output

[probonopd]

Currently the output files are stored along with the source files inside the git repository. This is less than ideal.

Would you be interested in having a CI workflow that would generate the output (e.g., EPUB) from the source files automatically each time something was committed to master?

This could be realized e.g., using Travis CI.
Let me know if you'd like me to look into it.

[jrbastien]

Yes, if you make sure all outputs are produced in every languages and that you find ways to adjust copyright dates automatically. The dates are on jpeg produced from an SVG template.

Also, I remember when I setup this procedure a few years ago that there was some manual adjustment needed to the PDF outputs before checking them back to GIT.

[probonopd]

I would not check the output files into the git repository at all, but rather deploy them to the website (for HTML) and/or upload them to GitHub Releases (for EPUB, PDF etc.). Do you think this would work for this project?

[jrbastien]

Yes, I did not mean to check in the output files but I meant that if you use the generation scripts provided, they create PDFs that are not deployment ready.

[probonopd]

adjust copyright dates automatically

Which copyright dates (where) do you mean? Thanks.

[probonopd]

So here is the first iteration, please do review carefully:
https://github.com/probonopd/designwithfontforge.com/releases/tag/continuous

This is how they are generated:
https://github.com/probonopd/designwithfontforge.com/blob/travis/.travis.yml

[probonopd]

Maybe implement this?

[jrbastien]

Not yet please. I just checked the PDFs. The overall process in Planning your project is a grey box and there is another grey box showing the page before. Also if you compare the actual version with the new one you generated, you will see that there is no image of the video in General Introduction.

Notice I'm not surprised, these were the things that I needed to manually fixed.

For the copyright, I'm referring to the cover page. We should see 2020. We could improved this by amending the svg template with variable that would get the timestamp and then generating the jpgs.

And lastly, the file is 27% bigger than before. But it could be because someone added decoration images in the first chapter without optimizing them.

[probonopd]

Thanks for reviewing the output.

Notice I'm not surprised, these were the things that I needed to manually fixed.

How do you fix something manually in an automatically generated PDF? Did you actually use an editor and literally edited the PDF by hand? Or did you do something manual prior to running ./create_books.sh?

[skef]

This is "not the place" but since multiple relevant parties are in discussion here I want to point out that my replacement Expand Stroke document wound up more or less buried among the old docs here: https://fontforge.org/docs/techref/stroke.html Some of the latter parts are probably too technical but the initial portion might be a good addition to designwithfontforge.

I also have a script to generate SVG figures like those in that doc, which are both more compact than images and don't have scaling issues. It takes an SFD, char, and scale factor as input.

[jrbastien]

I edited the books with a PDF editor after generating them with the script. The cover page creation could definitely be added to the script. We would then get the proper date.

For the other problems, I never took the time to analyze what could be done to prevent them upfront. If you have time to do this, that would be great.

And @skef, your comment is really out of place!

[probonopd]

@jrbastien yes I think the changes should be made prior to running ./create_books.sh. I will see what I can do.

[skef]

@jrbastien FontForge doesn't currently have reliable communication channels for this sort of thing. If you actually want me to open an issue about this, tell me so.

[probonopd]

@skef

I also have a script to generate SVG figures like those in that doc, which are both more compact than images and don't have scaling issues. It takes an SFD, char, and scale factor as input.

Can you show me how to use it?

[skef]

@probonopd Sure -- I'll get a copy into a repo and point you to it.