Ditch PDF Assembler and use PDF.js directly

[mrtcode]

PDF.js recently added a support for saving AcroForms annotations, and this serves as a very good example how to implement a basic append based PDF writing.

Append based writing fits probably better in our case as it's less invasive on PDF structure especially because annotations writing/deleting/modifying is relatively basic operation.

Therefore PDF Assembler becomes a useless intermediary that has to be maintained and reduces performance for this specific annotations-related use case. It also doesn't support encryption.

[chriscz]

Would this migration still support annotations made in other programs?

Suppose I created a single sample PDF with some text and images and then annotated this PDF using a variety of apps across several platforms like Android, Apple, iOS, Linux, Web, and Windows. These files could then be used to develop support for parsing a broad range of annotations. Do you think such a data sample would be useful, and maybe best submitted to pdfjs?