sm-discuss AT lists.ibiblio.org
Subject: Public SourceMage Discussion List
List archive
Re: [SM-Discuss] Should we use UPSTREAM_GPG to verify sources for end-users (cast)? I say no.
- From: Eric Sandall <eric AT sandall.us>
- To: sm-discuss AT lists.ibiblio.org
- Subject: Re: [SM-Discuss] Should we use UPSTREAM_GPG to verify sources for end-users (cast)? I say no.
- Date: Fri, 6 Mar 2020 10:19:07 -0800
On 2/28/20 07:14, Thomas Orgis wrote:
Am Thu, 27 Feb 2020 11:06:15 +0100 schrieb Thomas Orgis <thomas-forum AT orgis.org>:Upstream signatures are there for the guru to verify and establish trustFor that purpose, I'd really like the gpg files from the grimoire to be moved to a spell (the files themselves moved into an archive to be summoned) which installs something like /usr/share/upstream-keys/*.gpg . That way, they're actually more easy to use for interactive verification, with a known location, and we get rid of the binary blobs in the grimoire itself. That's 425 files with about 4M size total. The're a bit compressible, so the net effect on grimoire.tar.bz2 is from 13M to 10M if we leave them out. Thinking about this … a central collection of the “right” public keys to verify open source releases with would be useful in its own right. A kind of trusted versioned keyserver (database archive with known hash). Of course it would've been nice if the Web Of Trust had worked out, but I don't see that happening. I myself am also guilty of not seing the point in attending a key-signing event even when I was in the vicinity. At least for the Source Mage project, an smgl-upstream-keys spell could provide the same level of trust we have right now to the keys collected in the grimoire. I like the idea of separating them into a spell even if we keep on-cast GPG verificaton in combination with hash checking. We could add support for the admin to add to /usr/share/uptream-keys/local.gpg (or /usr/local/share/upstream-keys/add.gpg, whatever) to have keys marked as trusted for anything (even self-signing all sources once, for some strange reason). Alrighty then, Thomas
Those are all great points. :) I believe the UPSTREAM_GPG_KEY was when we assumed that upstream would always be a trusted source. It also avoided the issue when a tarball was re-relased without a new version tag, but was still properly signed, then we didn't have to do our own diff of the source code, do some sleuthing, and then update the SOURCE_HASH ourselves so that further casts would work. They're also convenient as we don't need to generate a hash on upgrades. ;)
I agree with the overhead and us keeping track and verifying the GPG trustworthiness. I don't see much of a downside for going back to SOURCE_HASH only (some SMGL devs still use SOURCE_HASH instead of GPG even when available). This would also avoid the issue of when gnupg breaks (e.g. some upgrades) and we can no longer cast spells with source validation until we fix it.
-Eric
Attachment:
signature.asc
Description: OpenPGP digital signature
-
Re: [SM-Discuss] Should we use UPSTREAM_GPG to verify sources for end-users (cast)? I say no.,
Eric Sandall, 03/06/2020
-
Re: [SM-Discuss] Should we use UPSTREAM_GPG to verify sources for end-users (cast)? I say no.,
Ismael Luceno, 03/16/2020
-
Re: [SM-Discuss] Should we use UPSTREAM_GPG to verify sources for end-users (cast)? I say no.,
Thomas Orgis, 03/16/2020
- Re: [SM-Discuss] Should we use UPSTREAM_GPG to verify sources for end-users (cast)? I say no., Ismael Luceno, 03/16/2020
-
Re: [SM-Discuss] Should we use UPSTREAM_GPG to verify sources for end-users (cast)? I say no.,
Thomas Orgis, 03/16/2020
-
Re: [SM-Discuss] Should we use UPSTREAM_GPG to verify sources for end-users (cast)? I say no.,
Ismael Luceno, 03/16/2020
Archive powered by MHonArc 2.6.24.