Changelog History
Page 4
-
v1.5.4 Changes
February 27, 2021- ๐ Undo prior retraction in v1.5.3 as we prepare for v5.0.0 release
- ๐ History of changes: see https://github.com/go-chi/chi/compare/v1.5.3...v1.5.4
-
v1.5.3 Changes
February 21, 2021- โก๏ธ Update go.mod to go 1.16 with new retract directive marking all versions without prior go.mod support
- ๐ History of changes: see https://github.com/go-chi/chi/compare/v1.5.2...v1.5.3
-
v1.5.2 Changes
February 10, 2021- โช Reverting allocation optimization as a precaution as go test -race fails.
- ๐ Minor improvements, see history below
- ๐ History of changes: see https://github.com/go-chi/chi/compare/v1.5.1...v1.5.2
-
v1.5.1 Changes
December 06, 2020- ๐ Performance improvement: removing 1 allocation by foregoing context.WithValue, thank you @bouk for your contribution (https://github.com/go-chi/chi/pull/555). Note: new benchmarks posted in README.
middleware.CleanPath
: new middleware that clean's request path of double slashes- ๐ deprecate & remove
chi.ServerBaseContext
in favour of stdlibhttp.Server#BaseContext
- ๐ plus other tiny improvements, see full commit history below
- ๐ History of changes: see https://github.com/go-chi/chi/compare/v4.1.2...v1.5.1
-
v1.5.0 Changes
November 12, 2020chi
dates back to 2016 with it's original implementation as one of the first routers to adopt the newly introduced context.Context api to the stdlib -- set out to design a router that is faster, more modular and simpler than anything else out there -- while not introducing any custom handler types or dependencies. Today,chi
still has zero dependencies, and in many ways is future proofed from changes, given it's minimal nature. Between versions, chi's iterations have been very incremental, with the architecture and api being the same today as it was originally designed in 2016. For this reason it ๐ makes chi a pretty easy project to maintain, as well thanks to the many amazing community contributions over the years ๐ to who all help make chi better (total of 86 contributors to date -- thanks all!).๐ Chi has been an labour of love, art and engineering, with the goals to offer beautiful ergonomics, flexibility, performance ๐ and simplicity when building HTTP services with Go. I've strived to keep the router very minimal in surface area / code size, ๐ฆ and always improving the code wherever possible -- and as of today the
chi
package is just 1082 lines of code (not counting middlewares, which are all optional). As well, I don't have the exact metrics, but from my analysis and email exchanges from ๐ companies and developers, chi is used by thousands of projects around the world -- thank you all as there is no better form of joy for me than to have art I had started be helpful and enjoyed by others. And of course I use chi in all of my own projects too :)๐ For me, the asthetics of chi's code and usage are very important. With the introduction of Go's module support (which I'm a big fan of), chi's past versioning scheme choice to v2, v3 and v4 would mean I'd require the import path of "github.com/go-chi/chi/v4", leading to the lengthy discussion at https://github.com/go-chi/chi/issues/462. Haha, to some, you may be scratching your head why I've spent > 1 year stalling to adopt "/vXX" convention in the import path -- which isn't horrible in general -- but for chi, I'm unable to accept it as I strive for perfection in it's API design, ๐ aesthetics and simplicity. It just doesn't feel good to me given chi's simple nature -- I do not foresee a "v5" or "v6", โฌ๏ธ and upgrading between versions in the future will also be just incremental.
I do understand versioning is a part of the API design as well, which is why the solution for a while has been to "do nothing", ๐ as Go supports both old and new import paths with/out go.mod. However, now that Go module support has had time to iron out kinks and is adopted everywhere, it's time for chi to get with the times. Luckily, I've discovered a path forward that will make me happy, ๐ while also not breaking anyone's app who adopted a prior versioning from tags in v2/v3/v4. I've made an experimental release of โ v1.5.0 with go.mod silently, and tested it with new and old projects, to ensure the developer experience is preserved, and it's โ largely unnoticed. Fortunately, Go's toolchain will check the tags of a repo and consider the "latest" tag the one with go.mod. However, you can still request a specific older tag such as v4.1.2, and everything will "just work". But new users can just โ
go get github.com/go-chi/chi
orgo get github.com/go-chi/chi@latest
and they will get the latest version which contains ๐ go.mod support, which is v1.5.0+.chi
will not change very much over the years, just like it hasn't changed much from 4 years ago. ๐ Therefore, we will stay on v1.x from here on, starting from v1.5.0. Any breaking changes will bump a "minor" release and ๐ backwards-compatible improvements/fixes will bump a "tiny" release.โฌ๏ธ For existing projects who want to upgrade to the latest go.mod version, run:
go get -u github.com/go-chi/[email protected]
, which will get you on the go.mod version line (as Go's mod cache may still remember v4.x). Brand new systems can run โgo get -u github.com/go-chi/chi
orgo get -u github.com/go-chi/chi@latest
to install chi, which will install v1.5.0+ ๐ built with go.mod support.๐ My apologies to the developers who will disagree with the decisions above, but, hope you'll try it and see it's a very minor request which is backwards compatible and won't break your existing installations.
Cheers all, happy coding!
-
v1.0.0 Changes
July 01, 2016- ๐ Released chi v1 stable https://github.com/go-chi/chi/tree/v1.0.0 for Go 1.6 and older.
-
v0.9.0 Changes
March 31, 2016- ๐ Reuse context objects via sync.Pool for zero-allocation routing #33
- ๐ฅ BREAKING NOTE: due to subtle API changes, previously
chi.URLParams(ctx)["id"]
used to access url parameters has changed to:chi.URLParam(ctx, "id")