PDFsharp & MigraDoc Foundation

I am using PdfSharp (installed with nuget) to try and make an extension to the awesome ApprovalTests.Net library.

I have two otherwise completely identical pdfs (I checked with winmerge) but they always have different trailer IDs.

This messes up ApprovalTests.

What goes into computing this field and can I have it be consistent?

PDF files contain date and time of their creation. Obviously this differs with every new run.

PDFsharp uses GUIDs for several items (including the trailer ID). These will be different with every new run.

I agree it would be nice to have a special mode for automatic testing that will create identical files. I will discuss this with Stefan.

_________________
Regards
Thomas Hoevel
PDFsharp Team

Thanks Thomas,

In the meantime I am doing this. Maybe me butchering pdf files will provide motivation

Any input from you guys as far as edge cases this might not catch is appreciated.

I don't have a better idea for a quick solution. I had a similar idea.

My boss thinks about creating images from PDF pages and comparing those images (by creating a new image with the differences between the two source images for human analysis).
This will be a sophisticated solution, but not a quick solution.

As a short-term solution, keep on butchering.

_________________
Regards
Thomas Hoevel
PDFsharp Team

Unfortunately creating images would limit the solution's application. ApprovalTests provides various reporters, you could reasonably use winmerge or a pdf comparer which would give far more useful info than images. Additionally, approved files with this framework are committed to source control - imagine even a simple 70 page pdf rasterized, imagine how large it would be.

But yes, in the meantime here we go.

Ideally, pdfsharp would have an option where you can provide your own method to generate an ID

Same question. Do you have any ideas for it now? Hope you can share more about your experience.

Best regards.