- From: Chris Lilley <chris@w3.org>
- Date: Sat, 1 Jul 2023 17:55:36 +0300
- To: "public-png@w3.org" <public-png@w3.org>
- Message-ID: <6141112e-635e-503e-e6ef-f797ce6b39fb@w3.org>
-------- Forwarded Message -------- Subject: [Moderator Action] Puzzling of PNG Error Handling Date: Thu, 29 Jun 2023 19:30:37 +0000 From: DLiang Fun <dliangfun@gmail.com> To: public-png@w3.org Dear PNG Working Group, I hope this email finds you well. I wanted to bring to your attention a minor matter regarding the PNG specification on error handling. 1) In regards to chunk types, the specification clearly states that each byte in the chunk type should be limited to ISO 646 letters (A-Z and a-z). Therefore, it appears reasonable to report errors if this condition is not satisfied. 2) The Error Handling section explicitly states that encountering an unknown ancillary chunk is never considered an error. Now, here's the silly question that arises: If a chunk type contains non-ISO 646 letters but is still an unknown ancillary chunk, should we report an error? While reading the specification, I do have the feeling that we should not report an error in such a case. However, the specification does not explicitly indicate that, in case of conflicts, we should follow the Error Handling philosophy first. I have observed divergent implementations regarding this matter. Specifically, while libspng and lodepng choose to ignore the chunk in question, libpng halts the parsing process and reports a fatal error. Hence, I believe it would be beneficial to explicitly clarify this aspect in the specification. I kindly request your consideration and input on this topic. Your guidance in ensuring consistency and clarity within the specification would be greatly appreciated. Thank you for your time and attention.. I am looking forward to hearing your insights on this matter. Best regards, Dliangfun
Received on Saturday, 1 July 2023 14:55:42 UTC