Open jelmer opened 1 month ago
I've created https://gitlab.com/perm.pub/dulwich-sshsig-union for testing, coordination, and a bit of experimentation.
I somehow failed to recognize that my goal of switching hidos from GitPython to Dulwich will also require signing in addition verification. :facepalm: So the mission of this thread has expanded to include signing too. But the code from grawity should make this relatively easy if I'm already doing all the work to test verification well.
Quick update on my thoughts, now that I've gotten verification partially working with hidos/sshsiglib/dulwich. I'm now planning to have hidos using Dulwich this month. But I think hidos is going to continue depending on GitPython & classic git for creating/amending/writing document succession for a long time. But reading document successions will be via Dulwich. So verification with Dulwich is high priority but signing commits via Dulwich is low priority for me for awhile. I still want to work towards getting signing into Dulwich too, but it's just not as urgent. I hope signing isn't that much work once verification is working.
@jelmer Here's a quick update on my current thinking. I'm thinking of releasing version 2.0 of the hidos package this month and including a "vendored" library of sshsig
inside the hidos
package, but not as a submodule of hidos
. sshsig
and hidos
are separate top level modules. Then as a follow-on stage, some of the code can be copied and/or sshsig
can turn into a separate stand-alone package. Nobody that wants just the sshsig
functionality will also want hidos
.
sshsig
code is also currently in a stand alone repository, currently at https://gitlab.com/perm.pub/sshsiglib/. For now, this repo is also submodule inside inside the hidos
repository so that it can be packaged/vendored along with hidos
2.0.
I'll post some more details here this week on the sshsig
and dulwich
functions and types that hidos
is calling which hopefully can provide some insights and guidance on improved approaches in other dulwich applications.
Discussed in https://github.com/jelmer/dulwich/discussions/1391