EQS-512: Spike - Alternatives to pdfkit performance investigation

[liamtoozer]

What is the context of this PR?

The pdfkit package and wkhtmltopdf engine used in Runner to export HTML to PDF are now deprecated.

There was a Spike carried out about 3 years ago to look into moving to weasyprint. At the time, it was found that weasyprint was not as efficient as exporting to PDF as our current pdfkit. However, things have since moved on for weasyprint, and the deprecation for pdfkit was put in place just 2 years ago, so may be worth another look.

Part of this spike also looks at the xhtml2pdf package as well

How to review

• Read through the doc/performance-investigations/0025-pdfkit-alternatives/summary.md. As mentioned in the previous Spike, think about the following:
  • Does the document accurately reflect the test?
  • Are the observations accurate?
  • Can any other observations be made?
  • Do you agree with the recommendations?
  • Should any other recommendations be made?

[farres1]

The write-up looks good and summarises the information clearly. If we can dedicate time to researching Option 4, I agree with your conclusion of Options 2, 4 and 7. Otherwise, I think combining Options 2 and 7 would be the most beneficial approach.

[Yuyuutsu]

Really nice investigation!
I think the next step we choose is really depends on the resources we have. I would like us to look into our own PDF generation but that is a monumental task. Even changing to weasyprint, which looks like our best option, is still a really big task. I do like the idea of incorporating a loop icon for better UX.

This is most likely something we may need to discuss as a whole team.