3 ms·

It should be said that PDF signatures are a very fragile design, leading to a pletora of security issues: https://pdf-insecurity.org/ https://pdf-insecurity.org

by hannob 4y ago

It should be said that PDF signatures are a very fragile design, leading to a pletora of security issues: https://pdf-insecurity.org/ https://pdf-insecurity.org/ A core problem is that a pdf signature does not necessarily cover a complete file, but can be a partial signature. This adds a whole lot of complexity and unclarity around what is actually signed, allowing all kinds of attacks. I feel this is all so problematic that if you want to sign PDFs it's probably better to not use PDF signatures, but some form of outside signatures over the whole file.

Avamander 4y ago
It is in theory possible to craft better-signed PDFs that would pass the European eIDAS Qualified Electronic Signature requirements. But in general the ASiC-E container format is more versatile and also more robust against potential flaws.
- kjetil 4y ago
  In practice, though, PAdES has a lot more support and has the crucial property of being easy to view by end-users. Is there any wide use of ASiC?
  - Avamander 4y ago
    > In practice, though, PAdES has a lot more support and has the crucial property of being easy to view by end-users. For now, I'd expect ever increasing compliance with eIDAS in the future. PDFz are also not the only thing that people want to sign, that's where an agnostic container format has its benefits. > Is there any wide use of ASiC? I know that at least Estonia, Latvia, Lithuania and Finland have deployed it. Of those Estonia probably has the widest and longest use of it, as they migrated *to* ASiC-E, having used the predecessors BDOC and CDOC previously.
layer8 4y ago
The problem is mostly (1) signature validation implementations that don’t perform all necessary checks, and (2) PDF viewers with signature validation that don’t properly guide the user to view the actually signed document revision (when it is not the most recent one). Regarding (1), this is because there exists no complete specification of what an implementation needs to check with regard to the PDF format. Implementors have to figure it out on their own, and thus there have been some gaps.

by hannob 4y ago

Avamander 4y ago

It is in theory possible to craft better-signed PDFs that would pass the European eIDAS Qualified Electronic Signature requirements. But in general the ASiC-E container format is more versatile and also more robust against potential flaws.

kjetil 4y ago
In practice, though, PAdES has a lot more support and has the crucial property of being easy to view by end-users. Is there any wide use of ASiC?
- Avamander 4y ago
  > In practice, though, PAdES has a lot more support and has the crucial property of being easy to view by end-users. For now, I'd expect ever increasing compliance with eIDAS in the future. PDFz are also not the only thing that people want to sign, that's where an agnostic container format has its benefits. > Is there any wide use of ASiC? I know that at least Estonia, Latvia, Lithuania and Finland have deployed it. Of those Estonia probably has the widest and longest use of it, as they migrated *to* ASiC-E, having used the predecessors BDOC and CDOC previously.

layer8 4y ago

The problem is mostly (1) signature validation implementations that don’t perform all necessary checks, and (2) PDF viewers with signature validation that don’t properly guide the user to view the actually signed document revision (when it is not the most recent one). Regarding (1), this is because there exists no complete specification of what an implementation needs to check with regard to the PDF format. Implementors have to figure it out on their own, and thus there have been some gaps.