Cache Revalidation#2570
Conversation
3862057 to
ab1df20
Compare
0d5fe20 to
c45ced3
Compare
|
I tested it and it doesn't work. Wrong results + crash. |
5ab805c to
fd8cb46
Compare
fd8cb46 to
2414a67
Compare
|
@mookums |
|
I think we should apply immediately https://developer.mozilla.org/en-US/docs/Web/HTTP/Guides/Caching#heuristic_caching But it may be challenging to parse the |
58c37f7 to
f4db7c7
Compare
|
Duration regression is from the cache tests being added to puppeteer. Any objections to just upping the average time in the CI test? |
|
Duration is about |
krichprollsch
left a comment
There was a problem hiding this comment.
Here is another comment from Claude about DeferredLayer (lines are outdated)
Not sure if it's correct r not 🤔
- DeferringLayer doesn't propagate .handled — its deferred-replay paths (DeferringLayer.zig:283- and :333-) only branch on .abort vs not, treating .handled like .proceed. If a CacheLayer 304 (.handled) ever sits inside a DeferringLayer, the replay would forward header and continue replaying data/done, double-forwarding. Likely not a current combination, but the asymmetry is a latent trap.
|
I think we are good to merge. Anything to change from your POV @karlseguin? |
|
@krichprollsch LGTM |
This adds Cache Revalidation, properly checking
no-cacheand stale entries.