Caching #480

Merged
Plume_migration_agent merged 3 commits from cache into master 5 years ago
Owner

Try to cache static resources.
Static non media file's timestamps are hashed at compile tome so that modifying them invalidate old cache.
For some reason, plume-front.wasm isn't loaded from cache, even if it contains the right headers.
I changed nothing and now it is? I guess I'm ok with it
I'll look into adding Etag support too
All response made from templates and not containing a form are now Etag-cached

Try to cache static resources. Static non media file's timestamps are hashed at compile tome so that modifying them invalidate old cache. ~~For some reason, plume-front.wasm isn't loaded from cache, even if it contains the right headers.~~ I changed nothing and now it is? I guess I'm ok with it ~~I'll look into adding Etag support too~~ All response made from templates and not containing a form are now Etag-cached
igalic (Migrated from github.com) reviewed 5 years ago
igalic (Migrated from github.com) left a comment

👀

👀
@ -5,0 +13,4 @@
.spawn()
.expect("failed find command");
let sort = Command::new("sort")
igalic (Migrated from github.com) commented 5 years ago

i'd rather this wasn't random, but deterministic

such as the file's she256sum

i'd rather this wasn't random, but deterministic such as the file's she256sum
trinity-1686a reviewed 5 years ago
@ -5,0 +13,4 @@
.spawn()
.expect("failed find command");
let sort = Command::new("sort")
Poster
Owner

What is your issue with it being random? It's fixed at compile time, so assuming a production server, which may restart but won't recompile often, the cache will be ok. My first through was using git commit sha, but this may be unnecessary information disclosure

What is your issue with it being random? It's fixed at compile time, so assuming a production server, which may restart but won't recompile often, the cache will be ok. My first through was using git commit sha, but this may be unnecessary information disclosure
igalic (Migrated from github.com) reviewed 5 years ago
@ -5,0 +13,4 @@
.spawn()
.expect("failed find command");
let sort = Command::new("sort")
igalic (Migrated from github.com) commented 5 years ago

if nothing about the code changes, the cache_id shouldn't change either

if nothing about the code changes, the cache_id shouldn't change either
codecov[bot] commented 5 years ago (Migrated from github.com)

Codecov Report

Merging #480 into master will decrease coverage by 0.31%.
The diff coverage is 0%.

@@            Coverage Diff             @@
##           master     #480      +/-   ##
==========================================
- Coverage   27.13%   26.82%   -0.32%     
==========================================
  Files          64       64              
  Lines        7337     7423      +86     
==========================================
  Hits         1991     1991              
- Misses       5346     5432      +86
# [Codecov](https://codecov.io/gh/Plume-org/Plume/pull/480?src=pr&el=h1) Report > Merging [#480](https://codecov.io/gh/Plume-org/Plume/pull/480?src=pr&el=desc) into [master](https://codecov.io/gh/Plume-org/Plume/commit/90baf9beb1884a886d4f2bf9df96f761c582d765?src=pr&el=desc) will **decrease** coverage by `0.31%`. > The diff coverage is `0%`. ```diff @@ Coverage Diff @@ ## master #480 +/- ## ========================================== - Coverage 27.13% 26.82% -0.32% ========================================== Files 64 64 Lines 7337 7423 +86 ========================================== Hits 1991 1991 - Misses 5346 5432 +86 ```
elegaanz (Migrated from github.com) reviewed 5 years ago
@ -5,0 +13,4 @@
.spawn()
.expect("failed find command");
let sort = Command::new("sort")
elegaanz (Migrated from github.com) commented 5 years ago

The build won't be run again if the code didn't changed, so the ID will not change.

The build won't be run again if the code didn't changed, so the ID will not change.
elegaanz (Migrated from github.com) reviewed 5 years ago
elegaanz (Migrated from github.com) commented 5 years ago

You should probably remove it from Cargo.toml too.

You should probably remove it from `Cargo.toml` too.
trinity-1686a reviewed 5 years ago
Poster
Owner

I totally forgot about that

I totally forgot about that
igalic (Migrated from github.com) reviewed 5 years ago
@ -5,0 +13,4 @@
.spawn()
.expect("failed find command");
let sort = Command::new("sort")
igalic (Migrated from github.com) commented 5 years ago

yes, in our development environment that is true, but not for anyone building packages or container images of our software. they always start from a clean slate

yes, in our development environment that is true, but not for anyone building packages or container images of our software. they always start from a clean slate
igalic (Migrated from github.com) reviewed 5 years ago
igalic (Migrated from github.com) left a comment

👍

👍
@ -5,0 +13,4 @@
.spawn()
.expect("failed find command");
let sort = Command::new("sort")
igalic (Migrated from github.com) commented 5 years ago

btw, is there no way to do this in rust?

btw, is there no way to do this in rust?
trinity-1686a reviewed 5 years ago
@ -5,0 +13,4 @@
.spawn()
.expect("failed find command");
let sort = Command::new("sort")
Poster
Owner

I think there are plenty, but I'm lazy and it's a build script. I prototyped using bash, so my rust code is basically bash translated (if you find it really bad, I can change that but I don't know if it's really useful)

I think there are plenty, but I'm lazy and it's a build script. I prototyped using bash, so my rust code is basically bash translated (if you find it really bad, I can change that but I don't know if it's really useful)
igalic (Migrated from github.com) reviewed 5 years ago
@ -5,0 +13,4 @@
.spawn()
.expect("failed find command");
let sort = Command::new("sort")
igalic (Migrated from github.com) commented 5 years ago

nah, i was just curious

nah, i was just curious
elegaanz (Migrated from github.com) reviewed 5 years ago
@ -11,0 +37,4 @@
.merge(HtmlCt(self.0).respond_to(r)?)
.header(ETag(EntityTag::strong(etag)))
.ok()
}
elegaanz (Migrated from github.com) commented 5 years ago

Probably obvious, but why do we disable caching on pages with a form?

Probably obvious, but why do we disable caching on pages with a form?
trinity-1686a reviewed 5 years ago
@ -11,0 +37,4 @@
.merge(HtmlCt(self.0).respond_to(r)?)
.header(ETag(EntityTag::strong(etag)))
.ok()
}
Poster
Owner

If the page contain a form, the csrf token will be cached, and may no longer match user's cookies. As these tokens are added later by a fairing, there is no way to detect the page changed here. The other solution could be to allow caching of page and for rocket_csrf to remove ETag when appropriate

If the page contain a form, the csrf token will be cached, and may no longer match user's cookies. As these tokens are added later by a fairing, there is no way to detect the page changed here. The other solution could be to allow caching of page and for rocket_csrf to remove ETag when appropriate
elegaanz (Migrated from github.com) approved these changes 5 years ago
elegaanz (Migrated from github.com) left a comment

👍 Thanks!

:+1: Thanks!

Reviewers

The pull request has been merged as a2b2e37aa0.
You can also view command line instructions.

Step 1:

From your project repository, check out a new branch and test the changes.
git checkout -b cache master
git pull origin cache

Step 2:

Merge the changes and update on Forgejo.
git checkout master
git merge --no-ff cache
git push origin master
Sign in to join this conversation.
No reviewers
No Milestone
No Assignees
2 Participants
Notifications
Due Date
The due date is invalid or out of range. Please use the format 'yyyy-mm-dd'.

No due date set.

Dependencies

No dependencies set.

Reference: Plume/Plume#480
Loading…
There is no content yet.