syn2mas: Support migrating external IDs as upstream OAuth2 providers #3917
Conversation
Deploying matrix-authentication-service-docs with
|
| Latest commit: |
6b02497
|
| Status: | ✅ Deploy successful! |
| Preview URL: | https://6f583dfb.matrix-authentication-service-docs.pages.dev |
| Branch Preview URL: | https://rei-syn2mas-7-external-ids.matrix-authentication-service-docs.pages.dev |
reivilibre
force-pushed
the
rei/syn2mas_7_external_ids
branch
from
704ddb7 to
6b02497
Compare
| #[allow(clippy::missing_panics_doc)] // not a real panic | ||
| #[tracing::instrument(skip_all, level = Level::DEBUG)] | ||
| pub async fn write_users(&mut self, users: Vec<MasNewUser>) -> Result<(), Error> { | ||
| pub fn write_users(&mut self, users: Vec<MasNewUser>) -> BoxFuture<'_, Result<(), Error>> { |
There was a problem hiding this comment.
What's the reasoning with boxing here, and why not -> impl Future?
| let mut link_ids: Vec<Uuid> = Vec::with_capacity(links.len()); | ||
| let mut user_ids: Vec<Uuid> = Vec::with_capacity(links.len()); | ||
| let mut upstream_provider_ids: Vec<Uuid> = Vec::with_capacity(links.len()); | ||
| let mut subjects: Vec<String> = Vec::with_capacity(links.len()); | ||
| let mut created_ats: Vec<DateTime<Utc>> = Vec::with_capacity(links.len()); | ||
|
|
||
| for MasNewUpstreamOauthLink { | ||
| link_id, | ||
| user_id, | ||
| upstream_provider_id, | ||
| subject, | ||
| created_at, | ||
| } in links | ||
| { | ||
| link_ids.push(link_id); | ||
| user_ids.push(user_id); | ||
| upstream_provider_ids.push(upstream_provider_id); | ||
| subjects.push(subject); | ||
| created_ats.push(created_at); | ||
| } |
There was a problem hiding this comment.
itertools::multiunzip is a thing, not sure it would make that better though so 🤷
There was a problem hiding this comment.
mrmm, it's hard to find a tool that works here other than doing it the boring way. I don't see that much point in destructuring to a tuple. Maybe it's a little better, but not by much
There was a problem hiding this comment.
posterity note: the crates soa_derive and columnar look like they could be promising for this sort of thing, but it doesn't seem worth adding a new dep (+ proc_macro + compile time) for this when we only have a handful of these to do
| type WriteBufferFlusher<'conn, T> = | ||
| for<'a> fn(&'a mut MasWriter<'conn>, Vec<T>) -> BoxFuture<'a, Result<(), Error>>; |
There was a problem hiding this comment.
Riiight, all the box future make sense now :)
| rng: &mut impl RngCore, | ||
| ) -> Result<UsersMigrated, Error> { | ||
| let mut write_buffer = MasUserWriteBuffer::new(mas); | ||
| let mut user_buffer = MasWriteBuffer::new(MasWriter::write_users); |
There was a problem hiding this comment.
not sure how I feel about passing function pointers instead of using traits… I guess that works, not sure how it would look like with traits
sandhose
added
A-Migration
Commit-by-commit suggested