Scripts are creating pathnames with trailing dashes

[selfissued]

Specs are created with dashes before the extensions in the pathnames when Implementer's Drafts are published. My local instances of these before I removed them with "git rm" and then the "OIDF Sync" action were:

C:\Users\micha\Documents\GitHub\openid\publication\sync\specs>dir *-.*

 Directory of C:\Users\micha\Documents\GitHub\openid\publication\sync\specs

11/20/2025  02:26 PM           101,815 openid-connect-native-sso-1_0-.html
11/20/2025  02:26 PM            37,492 openid-connect-native-sso-1_0-.txt
11/20/2025  02:26 PM            42,670 openid-connect-native-sso-1_0-.xml
09/19/2025  11:38 AM            98,163 openid-connect-rp-metadata-choices-1_0-.html
09/19/2025  11:38 AM            32,439 openid-connect-rp-metadata-choices-1_0-.txt
09/19/2025  11:38 AM            44,698 openid-connect-rp-metadata-choices-1_0-.xml

These correspond to these published specs:

C:\Users\micha\Documents\GitHub\openid\publication\sync\specs>dir openid-connect-native-sso-1_0-ID2*

 Directory of C:\Users\micha\Documents\GitHub\openid\publication\sync\specs

11/20/2025  02:26 PM           101,815 openid-connect-native-sso-1_0-ID2.html
11/20/2025  02:26 PM            37,492 openid-connect-native-sso-1_0-ID2.txt
11/20/2025  02:26 PM            42,670 openid-connect-native-sso-1_0-ID2.xml

and

C:\Users\micha\Documents\GitHub\openid\publication\sync\specs>dir openid-connect-rp-metadata-choices-1_0-ID1.*

 Directory of C:\Users\micha\Documents\GitHub\openid\publication\sync\specs

09/19/2025  11:38 AM            98,163 openid-connect-rp-metadata-choices-1_0-ID1.html
09/19/2025  11:38 AM            32,439 openid-connect-rp-metadata-choices-1_0-ID1.txt
09/19/2025  11:38 AM            44,698 openid-connect-rp-metadata-choices-1_0-ID1.xml

These specs should be published, but without the extra dashes at the ends of the names.

[jogu]

I think this should be resolved with https://github.com/openid/publication-checks/pull/3/changes

The actual root problem is that https://openid.net/specs/openid-connect-rp-metadata-choices-1_0-ID1.html has the wrong title - it says 'draft' instead of 'implementers draft' causing it to go down a wrong code path. The changes about validate the filename (e.g. -ID1) and the title (e.g. implementers draft) are consistent.

[selfissued]

The actual root problem is that https://openid.net/specs/openid-connect-rp-metadata-choices-1_0-ID1.html has the wrong title - it says 'draft' instead of 'implementers draft' causing it to go down a wrong code path.

No place in our publication procedures to date does a draft title include "Implementer's Draft". Our naming conventions are documented at https://openid.net/wg/resources/naming-and-contents-of-specifications/. It says this about Implementer's Drafts:

Specifications that are Implementer’s Drafts are also published at openid.net/specs/ with filenames indicating the Implementer’s Draft number, such as openid-client-initiated-backchannel-authentication-core-1_0-ID2.html.

You can tell that a spec is an Implementer's Draft because the pathname includes -ID* - not because of the title. This was done intentionally so that a proposed implementers draft can be copied without changes to the -ID* pathname without changes upon approval.

I'd be open to a discussion among experts - probably the working group chairs - about changing our procedures to require editing proposed Implementer's Drafts to change the title and date before publication as an Implementer's Draft. But this shouldn't be done silently as code change to the publication tool. It should be done deliberately with broad review, and with the corresponding documentation being updated accordingly.

Does the code currently expect "Implementer's Draft" in the title of Implementer's Drafts being published?

[jogu]

Yeah, you're right - I might've misdiagnosed the cause. I'm pretty sure it went down a code path it shouldn't have, or I'm confused about other specs having worked okay.

Either way the incorrect regex that was resulting in the - not being removed is fixed in https://github.com/openid/publication-checks/pull/3/changes so that particular problem shouldn't happen again.