HCIP: 1
Title: HCIP Purpose and Guidelines
Status: Draft
Type: Meta
Author: Your Name <your.email@domain.com>
Created: 2019-04-20
HCIP stands for Heterogeneous Computing Improvement Proposal. Each HCIP is a design document providing information to the OHCF community, or describing a new feature for Heterogeneous Computing or its processes or environment. The HCIP should provide a concise technical specification of the feature and a rationale for the feature. The HCIP author is responsible for building consensus within the community and documenting dissenting opinions.
We intend HCIPs to be the primary mechanisms for proposing new features, for collecting community input on an issue and for documenting the design decisions that have gone into Heterogeneous Computing. Because the HCIPs are maintained as text files in a versioned repository, their revision history is the historical record of the feature proposal.
There are three kinds of HCIP:
- A Standard Track HCIP describes any change that affects most or all Heterogeneous Computing implementations, which can be broken down into the following categories.
**HDK
**SDK
**Cloud Management
**Bus
**OS
**Application enablement
-
An Informational HCIP describes an Heterogeneous Computing design issue, or provides general guidelines or information to the OHCF community, but does not propose a new feature. Informational HCIPs do not necessarily represent the OHCF community consensus or a recommendation, so users and implementors are free to ignore Informational HCIPs or follow their advice.
-
A Meta HCIP describes a process surrounding Heterogeneous Computing or proposes a change to (or an event in) a process. Meta HCIPs are like Standard Track HCIPs but apply to areas other than the Heterogeneous Computing related development itself. They may propose an implementation, but not to Heterogeneous Computing's codebase; they often require community consensus; unlike Informational HCIPs, they are more than recommendations, and users are typically not free to ignore them. Examples include procedures, guidelines, changes to the decision-making process, and changes to the tools or environment used in Heterogeneous Computing development.
The HCIP repository Collaborators change the HCIPs status. Please send all HCIP-related email to the HCIP Collaborators, which are listed under HCIP Editors below. Also see HCIP Editor Responsibilities & Workflow.
The HCIP process begins with a new idea for Heterogeneous Computing. It is highly recommended that a single HCIP contain a single key proposal or new idea. Small enhancements or patches that don't affect consensus often don't need an HCIP and can be injected into the Heterogeneous Computing development workflow with a patch submission to the corresponding Heterogeneous Computing issue tracker. The more focused the HCIP, the more successful it tends to be. The HCIP Editor reserves the right to reject HCIP proposals if they appear too unfocused or too broad. If in doubt, split your HCIP into several well-focused ones.
Each HCIP must have a champion -- someone who writes the HCIP using the style and format described below, shepherds the discussions in the appropriate forums, and attempts to build community consensus around the idea. The HCIP champion (a.k.a. Author) should first attempt to ascertain whether the idea is HCIP-able. Opening an Issue is the best way to go about this.
Vetting an idea publicly before going as far as writing an HCIP is meant to save the potential author time. Asking the OHCF community first if an idea is original helps prevent too much time being spent on something that is guaranteed to be rejected based on prior discussions (searching the Internet does not always do the trick). It also helps to make sure the idea is applicable to the entire community and not just the author. Just because an idea sounds good to the author does not mean it will work for most people in most areas where Heterogeneous Computing is used.
Once the champion has asked the OHCF community as to whether an idea has any chance of acceptance, a draft HCIP should be presented as a Pull Request. This gives the author a chance to flesh out the draft HCIP; to make properly formatted, of high quality, and to address initial concerns about the proposal.
If the HCIP collaborators approves, the HCIP editor will assign the HCIP a number, label it as Standards Track, Informational, or Process, give it status "Draft", and add it to the git repository. The HCIP editor will not unreasonably deny an HCIP. Reasons for denying HCIP status include duplication of effort, being technically unsound, not providing proper motivation or addressing backwards compatibility, or not in keeping with the Heterogeneous Computing philosophy.
The HCIP author may update the Draft as necessary in the git repository. Updates to drafts may also be submitted by the author as pull requests.
Standards Track HCIPs consist of three parts, a design document, implementation and finally if warranted an update to the formal specification. The HCIP should be reviewed and accepted before an implementation is begun, unless an implementation will aid people in studying the HCIP. Standards Track HCIPs must be implemented in at least two viable Heterogeneous Computing clients before it can be considered Final.
HCIP authors are responsible for collecting community feedback on an HCIP before submitting it for review. However, wherever possible, long open-ended discussions should be avoided. Strategies to keep the discussions efficient include: having the HCIP author accept private comments in the early design phases, setting up a wiki page or git repository, etc. HCIP authors should use their discretion here.
For an HCIP to be accepted it must meet certain minimum criteria. It must be a clear and complete description of the proposed enhancement. The enhancement must represent a net improvement. The proposed implementation, if applicable, must be solid and must not complicate the protocol unduly.
Once an HCIP has been accepted, the implementations must be completed. When the implementation is complete in at least two viable clients and accepted by the community, the status will be changed to "Final". An update to the formal specification should accompany the "Final" status change.
An HCIP can also be assigned status "Deferred". The HCIP author or editor can assign the HCIP this status when no progress is being made on the HCIP. Once an HCIP is deferred, the HCIP editor can re-assign it to draft status.
An HCIP can also be "Rejected". Perhaps after all is said and done it was not a good idea. It is still important to have a record of this fact.
HCIPs can also be superseded by a different HCIP, rendering the original obsolete. This is intended for Informational HCIPs, where version 2 of an API can replace version 1.
The possible paths of the status of HCIPs are as follows:
Some Informational and Process HCIPs may also have a status of "Active" if they are never meant to be completed. E.g. HCIP 1 (this HCIP).
Each HCIP should have the following parts:
-
Preamble -- RFC 822 style headers containing metadata about the HCIP, including the HCIP number, a short descriptive title (limited to a maximum of 44 characters), the names, and optionally the contact info for each author, etc.
-
Abstract -- a short (~200 word) description of the technical issue being addressed.
-
Specification -- The technical specification should describe the syntax and semantics of any new feature. The specification should be detailed enough to allow competing, interoperable implementations for any of the current Heterogeneous Computing platforms (OCP, OpenStack, Kubernetes, RISC-V, ...).
-
Motivation -- The motivation is critical for HCIPs that want to change the Heterogeneous Computing related Implementations. It should clearly explain why the existing protocol specification is inadequate to address the problem that the HCIP solves. HCIP submissions without sufficient motivation may be rejected outright.
-
Rationale -- The rationale fleshes out the specification by describing what motivated the design and why particular design decisions were made. It should describe alternate designs that were considered and related work, e.g. how the feature is supported in other languages.
-
The rationale should provide evidence of consensus within the community and discuss important objections or concerns raised during discussion.
-
Backwards Compatibility -- All HCIPs that introduce backwards incompatibilities must include a section describing these incompatibilities and their severity. The HCIP must explain how the author proposes to deal with these incompatibilities. HCIP submissions without a sufficient backwards compatibility treatise may be rejected outright.
-
Implementations -- The implementations must be completed before any HCIP is given status "Final", but it need not be completed before the HCIP is accepted. It is better to finish the specification and rationale first and reach consensus on it before writing code.
HCIPs should be written in markdown format. Image files should be included in a subdirectory for that HCIP.
Each HCIP must begin with an RFC 822 style header preamble. The headers must appear in the following order. Headers marked with "* " are optional and are described below. All other headers are required.
HCIP: <HCIP number>
Title: <HCIP title>
Author: <list of authors' real names and optionally, email address>
* Discussions-To: <email address>
Status: <Draft | Active | Accepted | Deferred | Rejected |
Withdrawn | Final | Superseded>
Type: <Standards Track | Informational Meta>
Created: <date created on, in ISO 8601 (yyyy-mm-dd) format>
* Replaces: <HCIP number>
* Superseded-By: <HCIP number>
* Resolution: <url>
The Author header lists the names, and optionally the email addresses of all the authors/owners of the HCIP. The format of the Author header value must be
Random J. User address@dom.ain
if the email address is included, and just
Random J. User
if the address is not given.
If there are multiple authors, each should be on a separate line following RFC 2822 continuation line conventions.
Note: The Resolution header is required for Standards Track HCIPs only. It contains a URL that should point to an email message or other web resource where the pronouncement about the HCIP is made.
While an HCIP is in private discussions (usually during the initial Draft phase), a Discussions-To header will indicate the mailing list or URL where the HCIP is being discussed. No Discussions-To header is necessary if the HCIP is being discussed privately with the author.
The Type header specifies the type of HCIP: Standards Track, Informational, or Meta.
The Created header records the date that the HCIP was assigned a number. Both headers should be in yyyy-mm-dd format, e.g. 2001-08-14.
HCIPs may have a Requires header, indicating the HCIP numbers that this HCIP depends on.
HCIPs may also have a Superseded-By header indicating that an HCIP has been rendered obsolete by a later document; the value is the number of the HCIP that replaces the current document. The newer HCIP must have a Replaces header containing the number of the HCIP that it rendered obsolete.
HCIPs may include auxiliary files such as diagrams. Such files must be named HCIP-XXXX-Y.ext, where "XXXX" is the HCIP number, "Y" is a serial number (starting at 1), and "ext" is replaced by the actual file extension (e.g. "png").
It occasionally becomes necessary to transfer ownership of HCIPs to a new champion. In general, we'd like to retain the original author as a co-author of the transferred HCIP, but that's really up to the original author. A good reason to transfer ownership is because the original author no longer has the time or interest in updating it or following through with the HCIP process, or has fallen off the face of the 'net (i.e. is unreachable or not responding to email). A bad reason to transfer ownership is because you don't agree with the direction of the HCIP. We try to build consensus around a HCIP, but if that's not possible, you can always submit a competing HCIP.
If you are interested in assuming ownership of a HCIP, send a message asking to take over, addressed to both the original author and the HCIP editor. If the original author doesn't respond to email in a timely manner, the HCIP editor will make a unilateral decision (it's not like such decisions can't be reversed :).
The current HCIP editors are
- All champions identified in the OHCF github organization
For each new HCIP that comes in, an editor does the following:
- Read the HCIP to check if it is ready: sound and complete. The ideas must make technical sense, even if they don't seem likely to be accepted.
- The title should accurately describe the content.
- Edit the HCIP for language (spelling, grammar, sentence structure, etc.), markup, code style
If the HCIP isn't ready, the editor will send it back to the author for revision, with specific instructions.
Once the HCIP is ready for the repository, the HCIP editor will:
-
Assign an HCIP number (almost always just the next available number)
-
Accept the corresponding pull request
-
List the HCIP in [[README.md]]
-
Send email back to the HCIP author with next step.
Many HCIPs are written and maintained by developers with write access to the Heterogeneous Computing codebase. The HCIP editors monitor HCIP changes, and correct any structure, grammar, spelling, or markup mistakes we see.
The editors don't pass judgment on HCIPs. We merely do the administrative & editorial part. Except for times like this, there's relatively low volume.
For Heterogeneous Computing implementers, HCIPs are a convenient way to track the progress of their implementation. Ideally each implementation maintainer would list the HCIPs that they have implemented. This will give end users a convenient way to know the current status of a given implementation or library.
HCIPs are intended to replace the venerable etherpads which described the initial PoC (Proof of Concept) and strike a balance between ease of accessibility and trackablity.
Fork the HCIP repo. Write an HCIP using the HCIP markdown template and initiate a pull request.
This document, recently edited for the specific use of OHCF community, was derived heavily from [https://github.com/bitcoin/bips Bitcoin's BIP-0001] written by Amir Taaki which in turn was derived from [https://www.python.org/dev/peps/ Python's PEP-0001]. In many places text was simply copied and modified. Although the PEP-0001 text was written by Barry Warsaw, Jeremy Hylton, and David Goodger, they are not responsible for its use in the Heterogeneous Computing Improvement Process, and should not be bothered with technical questions specific to, Heterogeneous Computing, or the HCIP. Please direct all comments to the HCIP editors.