Re: Atomic patches and network interruptions

[Glenn Strauss <[e-mail]>]

On Mon, Feb 26, 2024 at 08:30:38PM +0100, Marius Kleidl wrote:
> Dear working group,
> 
> RFC 5789, which defines the PATCH method, places this requirement:
> 
> The PATCH method requests that a set of changes described in the request
> > entity be applied to the resource identified by the Request-URI. The set of
> > changes is represented in a format called a "patch document" identified by
> > a media type.
> > [...]
> > The server MUST apply the entire set of changes atomically and never
> > provide (e.g., in response to a GET during this operation) a partially
> > modified representation. If the entire patch document cannot be
> > successfully applied, then the server MUST NOT apply any of the changes.

This is very direct and very explicit.  "atomically" means exactly that.

> > If the entire patch document cannot be
> > successfully applied, then the server MUST NOT apply any of the changes.

The server MUST NOT apply an incomplete patch document.  No exceptions.

Among your options:
The client can send smaller patch documents if changes are independent.
  (recommended)
The server *application* handling the request could buffer and store the
  (partial) request body fragments, allowing for resumable uploads, and
  then apply the changes only once the entire patch document is received

> Or could this be allowed if the
> corresponding media types explicitly allows applying partial content?

More questionable: a PATCH format might be defined/extended with a new
  media-type defining a format that consists of multiple patch documents
  which may be applied independently.  The client would need some way(s)
  to query/validate which patch documents were applied.  Again, per the
  RFC, each patch document must be applied atomically; never partially.

Cheers, Glenn