Checked against the current Windows build. Where an article names a menu, a button or a message, that is the wording in the application, not a paraphrase.
The failed tab carries the reason, and nothing else records it
The document tab shows the file name followed by failed, and the sentence explaining it sits in the middle of the window where the pages would have been. It is worth reading rather than closing, because four quite different things arrive at the same place.
- “This file isn’t a readable PDF.” The bytes are not a PDF at all, or not one the engine can make sense of.
- “This PDF uses a security scheme we can’t open.” It is encrypted with a handler this version does not implement, and no password gets you past that.
- “The document reader stopped while reading this file.” The worker process died mid-read, which usually means damage it cannot get past. Nothing was changed and nothing was saved.
- “Windows would not let Creasepoint read this file.” A permissions problem, or another application holding the file open.
None of this goes to a log, because there is no log: the application does not keep a list of the documents you have opened. Copy the sentence before you close the tab if you will need it later.
A file can be called a PDF without being one
The extension is a claim, not a format. A page saved out of a browser, an error page that arrived instead of the document you asked for, a photograph renamed by somebody being helpful — all of them appear in the file picker and none of them is a PDF. A real one begins with the five bytes %PDF-, which you can see in any text editor without understanding a thing about the rest of the file.
Other readers are more forgiving than this one. Browsers in particular repair a broken cross-reference table quietly and show you something, so “it opens in Edge” is not evidence that the file is well formed. For a second opinion from the same engine, creasepoint info statement.pdf prints the size, the page count and the object count per page, and refuses with the same sentence the window used.
Protection is a question here, not a refusal
A protected document does not fail to open; it asks. The prompt is headed “This document is protected.”, and a wrong answer brings it back reading “That password did not open it.” rather than giving up. What the password buys is a decrypted copy held in memory. The file on disk is never written to, still needs its password, and anything you save from the session is a new file.
Documents carrying an owner password and no user password open with no prompt at all, and are meant to: most government forms, bank statements and publisher files are shaped that way. If instead you were told the security scheme cannot be opened, that is not a password problem and typing one will not help. The two passwords in a PDF covers both cases.
Some documents are declined on purpose
Every file this application opens is treated as hostile, because a PDF arrives by email or from a portal and nobody vouches for it. The bounds are refusals rather than quiet truncations. A file over 512 MB is turned away before the engine sees a byte. A document over 10,000 pages is refused with its own page count in the message. A page larger than 14,400 units is refused, because rendering cost grows with the square of that number, and so is a single page carrying more than 200,000 objects.
Time is bounded the same way. An operation that ought to be quick — describing a document, drawing a preview, writing metadata — is given 90 seconds, and slow work such as OCR or a compression search is given five minutes, after which the worker is stopped. A deliberate refusal names the number and the limit, which is how you tell it from a failure.
It opened, and the pages are blank
Three different things look like this. A page can genuinely be empty. A page can be damaged in a way that leaves it in the file but unreadable, which creasepoint info reports as “N page(s) could not be described - damaged, or past a safety limit. The rest of this document is fine.” Or the document is longer than the preview: the viewer draws the first 40 pages and stops, while whole-document operations still run over every page in the file.
A fourth case is not a fault at all. A page showing a picture of typed text is a scan, and there is nothing wrong with it — it holds no text objects, so the find box reports nothing and double-clicking does nothing. That symptom has its own article.
The two places worth looking next.
Read the matching guide if you want the reasoning rather than the fix, and check limitations before spending any longer on it — some of what people look for here is absent on purpose.