Merge pull request #3052 from ChristopherDavenport/cachingHeadersMiddleware #3052
Conversation
ChristopherDavenport
force-pushed
the
cachingHeadersMiddleware
branch
from
b058b92
to
6248986
Compare
server/src/main/scala/org/http4s/server/middleware/Caching.scala
Outdated
Show resolved
Hide resolved
| } | ||
| } | ||
|
|
||
| object Helpers { |
There was a problem hiding this comment.
Does this need to be public? If it does, I might consider moving the two helpers herein to the top-level. Things nested in Helpers or Utils or bits tend to not be discoverable.
There was a problem hiding this comment.
The thing I was hoping to achieve was to allow those who used caching to have a way to get the tools we use rather than recreate them, but only leave the middlewares exposed at the top level.
| case finite: FiniteDuration => finite | ||
| case _ => 315360000.seconds // 10 years | ||
| // Http1 caches do not respect max-age headers, so to work globally it is recommended | ||
| // to explicitly set an Expire which requires some time interval to work |
There was a problem hiding this comment.
Nice comment. I was getting ready to throw a fit about Duration about 20 lines above.
| } | ||
|
|
||
| def publicCache[G[_]: MonadError[*[_], Throwable]: Clock, F[_]]( | ||
| lifetime: Duration, |
There was a problem hiding this comment.
I feel like we more commonly put the Http alone in a parameter list, for partial application. But I got that feeling a long time ago, before we had all the G and what not. I'm not sure whether that's still a useful convention.
There was a problem hiding this comment.
We could put it as the first argument, but if we put it in the second like we generally do, then G and F cannot infer
There was a problem hiding this comment.
Drat. I think you're right. That's sad, because I like to think of these as partially applied up to the Http.
There was a problem hiding this comment.
Alternative is building it as partially applied class. So that we can say Duration -> (forall g, f . ((MonadError Throwable g), Clock g) => (http g f) -> (http g f))
There was a problem hiding this comment.
Ooh, that's an interesting idea, if the type inference still works well enough when folding them down into one.
| .map(Instant.ofEpochMilli(_)) | ||
| .flatMap(nowInstant => HttpDate.fromInstant(nowInstant).liftTo[G]) | ||
| .map(nowHttpDate => HDate(nowHttpDate)) | ||
| HttpDate.current[G].map(HDate(_)) |
There was a problem hiding this comment.
To our future selves: we've usually resisted putting convenience constructors directly on headers, but things like this make me wonder.
| } | ||
| } | ||
|
|
||
| object Helpers { |
| } | ||
|
|
||
| def publicCache[G[_]: MonadError[*[_], Throwable]: Clock, F[_]]( | ||
| lifetime: Duration, |
There was a problem hiding this comment.
Drat. I think you're right. That's sad, because I like to think of these as partially applied up to the Http.
ChristopherDavenport
changed the title
No description provided.