Re: [Tools-discuss] Tools resources that we use: bap

[Julian Reschke <[e-mail]>]

Am 05.03.2021 um 16:18 schrieb Paul Kyzivat:
> On 3/5/21 9:36 AM, Robert Sparks wrote:
>> Please drop notes to the tools-discuss list or directly to me as you
>> think of things. I'd like for you to catch them as you see you need them.
>
>> On 3/5/21 2:51 AM, Carsten Bormann wrote:
>>> On 2021-03-05, at 09:39, Stewart Bryant <[e-mail]>
>>> wrote:
>>>> To whom do we send a list of the tools resources that we use?
>>> We could start a discussion thread on tools-discuss.
>
> I use bap from time to time. Unfortunately AFAIK it is barely
> maintained. After I finished RFC7405 I asked Bill Fenner about
> incorporating support for it in bap. He was very hard to reach. I then
> implemented the necessary updates to bap and submitted a pull request.
> But it was never acted upon. I eventually just gave up.

The http-core editors have their own fork (which does 7405, but also
handles the HTTP #list production). See
<https://github.com/httpwg/http-core/tree/master/bap>.

> I'd really like to have better support for ABNF in the draft pipeline.
> Ideally ABNF checking would be an automatic part of the process.
> Unfortunately it is hard to do well. Much ABNF in drafts/RFCs has
> dependencies on ABNF in some other document, and those dependencies are
> stated informally so that it takes manual processing to prepare
> something that can be fully checked by bap. If/how to resolve this needs
> discussion and then probably some implementation work.

In HTTP (core), we declare all ABNF productions that we import as prose,
for instance, see
<https://greenbytes.de/tech/webdav/draft-ietf-httpbis-semantics-14.html#uri.references>.

With that, the ABNF at least is self-contained and can be checked
automatically (and we do that in our build process).

Best regards, Julian