[media-types] Re: [IANA #1458209] application/vnd.fi duswriter+zip registration request
"Murray S. Kucherawy" <[email protected]>
| Newsgroups | gmane.ietf.types |
|---|---|
| Message-ID | <CAL0qLwZr=+C5xYHbE74Qx9upX7E8O3D8K7qsYaCkLdL4y+OAUA@mail.gmail.com> |
Inline. -MSK On Mon, Aug 17, 2026 at 5:39 PM Amanda Baber via RT < [email protected]> wrote: > Name: Johannes Wilm > > Email: [email protected] > > Media type name: application > > Media subtype name: vnd.fiduswriter+zip > > Required parameters: N/A. > > Optional parameters: N/A. > > Encoding considerations: binary > > This media type is a ZIP container (indicated by the "+zip" > structured syntax suffix). The archive is transferred as binary data and > MUST NOT be encoded as text. In MIME e-mail contexts, transfer encoding > such as base64 is required. > I suggest saying "binary" here is sufficient, and this prose is redundant to what's written elsewhere, so it could be trimmed. In any event, the MUST/SHOULD language is appropriate for the published specification but not for the registration template. > Security considerations: The content is a ZIP archive that contains a > declarative JSON document > together with images and optional template assets. The security > considerations are therefore similar to those of other ZIP-based > container formats (application/zip, ODF, EPUB), with the additional > concerns of JSON processing: > I suggest referring to the specifications for ZIP and JSON, which are https://www.iana.org/assignments/media-types/application/zip for the former and RFC 8259 for the latter. > - Archive handling: The archive may contain compressed entries. > Implementations MUST bound the total uncompressed size and the number > of entries to mitigate decompression-bomb ("zip bomb") attacks, and > MUST NOT write archive entries outside a designated target directory to > avoid path-traversal ("zip slip") vulnerabilities. > As above, I suggest "MUST" and "MUST NOT" become "must" and "must not". Alternatively, just remove this if it's covered in the published specification. > - Entry name handling: Entry names MUST be treated as opaque identifiers > inside the archive. Applications MUST NOT use them as filesystem paths > without sanitisation, and MUST NOT allow absolute paths or ".." > segments to escape the extraction directory. > Same. > - JSON payloads: The archive contains JSON files (document.json, > images.json, bibliography.json and optional template metadata). > Implementations MUST parse them with a strict JSON parser, reject > malformed input, and ignore unknown or unexpected fields. There is no > scripting or executable content defined by this media type; the JSON is > purely declarative. > Same. > - Embedded resources: The archive may contain images and template assets. > These MUST be treated as untrusted binary data. Applications should not > execute, render with active content, or otherwise run anything contained > in the archive. > Same. > - Content-type confusion: Because the container is a generic ZIP file, a > malicious actor could disguise unrelated ZIP content as a Fidus Writer > document. Implementations MUST validate the "mimetype" entry and the > "filetype-version" entry before processing the archive and MUST NOT rely > on the filename extension alone. > Same. The rest is fine. _______________________________________________ media-types mailing list -- [email protected] To unsubscribe send an email to [email protected]