Prints wrong pages if specific numbers have been specified

Thank you for your bug report. Could you check whether the pages 59 and 60 are really the pages 59 and 60 of your document, i.e. does the number displayed in the window's top row match 59 when you navigate to your page 59? I'm asking this because many PDF documents do not correctly number their pages so that e.g. the "true" page 3 is numbered "1" in the document itself. Evince obviously does not care about printed page numbers but only about the number set in the document.

If this is not the problem, could you provide an example PDF showing this issue?

Oh, I've found out what is the actual problem.
Usually, I see at the numbers of the pages in the preview pane. Please check the attached screenshot.

The PDF file has its own page numbers sequence, started with C1, then two empty numbers, then 1, 2, 3 etc (got 3 extra pages). This sequence matches the printed numbers of the scanned pages.
So, when I enter a page number (say, 59) in the "Goto page" input box, I really go to the page 59 which matches the table of contents. But it's not the actual number of the page in the file (59+3=62).

The problem is that these numbers work in different way for the "Goto page" input box and for the Print dialogue.

Thanks.

Just a disclaimer of the screen shot. It has 3 windows:
1) The evince window
2) Print dialogue
3) Print preview

Now I understand the logic, but still think it's a usability issue.

oops, i'm sorry, explanation, not a disclaimer.

I see, that's the kind of issue I was expecting. It is a really difficult problem, though, evince wants to support two kinds of page numbers: The "table of contents" page number, so that if the document mentions "see page 59" you can enter this number in the input box and jump to the respective page. The "real" page number, so you can print every page of the document (in your example, the first three pages would otherwise not be printable. It's easy to overlook, but evince actually gives both numbers for every page (next to the "59" in your screenshot it says "62 sur 292" -- so the "real" page number is 62).
That said, I agree that the situation is not optimal. There is already a bug report about this issue, further comments should be added there (or, even better, to the linked upstream bug report): #114462 . I'm therefore marking this bug as a duplicate.

Sorry, the correct link is Bug #114462