proposal: net/http: automatically set Content-Length from Len method on Body

[jingyugao]

What version of Go are you using (go version)?

$ go version
go version go1.15 darwin/amd64

Does this issue reproduce with the latest release?

yes

What operating system and processor architecture are you using (go env)?

go env Output

$ go env
GO111MODULE="on"
GOARCH="amd64"
GOBIN=""
GOCACHE="/Users/gaojingyu/Library/Caches/go-build"
GOENV="/Users/gaojingyu/Library/Application Support/go/env"
GOEXE=""
GOFLAGS=""
GOHOSTARCH="amd64"
GOHOSTOS="darwin"
GONOPROXY="*.byted.org"
GONOSUMDB="*.byted.org"
GOOS="darwin"
GOPATH="/Users/gaojingyu/go"
GOPRIVATE="*.byted.org"
GOPROXY="https://go-mod-proxy.byted.org"
GOROOT="/usr/local/Cellar/go/1.13.6/libexec"
GOSUMDB="sum.golang.org"
GOTMPDIR=""
GOTOOLDIR="/usr/local/Cellar/go/1.13.6/libexec/pkg/tool/darwin_amd64"
GCCGO="gccgo"
AR="ar"
CC="clang"
CXX="clang++"
CGO_ENABLED="1"
GOMOD="/Users/gaojingyu/Tmp/go.mod"
CGO_CFLAGS="-g -O2"
CGO_CPPFLAGS=""
CGO_CXXFLAGS="-g -O2"
CGO_FFLAGS="-g -O2"
CGO_LDFLAGS="-g -O2"
PKG_CONFIG="pkg-config"
GOGCCFLAGS="-fPIC -m64 -pthread -fno-caret-diagnostics -Qunused-arguments -fmessage-length=0 -fdebug-prefix-map=/var/folders/r1/fgd1nc4s32zcy8myvl17np8r0000gp/T/go-build991311603=/tmp/go-build -gno-record-gcc-switches -fno-common"

What did you do?

Now in http.NewRequestWithContext
it check wether body type is a *bytes.Buffer,*bytes.Reader or *strings.Reader, then call Len() to get content-length.

What did you expect to see?

Check wether the body type implement interfae{Len()int}

[rsc]

We know that Len is the right choice for those three types.
There's no widespread "Len-er" interface in Go, though, so I am skeptical of calling it always.
Maybe on some types Len might be a number of records?

[jingyugao]

We know that Len is the right choice for those three types.
There's no widespread "Len-er" interface in Go, though, so I am skeptical of calling it always.
Maybe on some types Len might be a number of records?

How about check ContentLength is zero?
If ContentLength is zero and body implement interfae{Len()int},then assign Len() to ContentLength.

[rsc]

@jingyugao I certainly agree that if the caller has explicitly set ContentLength then it should not be replaced.

But that doesn't really address my objection. Content-Length is not a hint. It is a promise, and the code should not make a promise unless it is sure it can be kept. The code is sure for the three special-case types (which are very common). The code cannot be anywhere near as sure about any type that happens to have a Len() int method. Len is measured in elements, which does not always mean bytes. (For example, suppose io.MultiReader returned a concrete implementation with a Len method that returned the number of underlying readers.)

Is there some widespread body implementation you are concerned about handling?

[rsc]

Adding to the proposal minutes. Len() int still seems like not enough signal for me.

[bcmills]

In #16474 (comment), @ianlancetaylor observed:

It seems that the best way to use this technique effectively is for each magic method to mean exactly one thing. … If multiple functions support the same magic method, we wind up with an N-x-N crossbar: trying to use the magic method for multiple different purposes, when we have to know which purpose is intended to know whether it makes sense.

For this case I would share Russ's concern about Len indicating the length of some container in units other than bytes — perhaps runes, number of entries, or similar.

[rsc]

Based on the discussion above, this seems like a likely decline.

[rsc]

No change in consensus, so declined.