Archive · Week 8 · Feb 16 – 22, 2026from 5 items
The Go team focused on deprecating an rarely‑used HTTP/2 scheduler, adding convenient cloning helpers for URLs and form values, improving file debugging output, tightening vet checks for scanner and SQL row errors, and announcing a new go fix implementation in Go 1.26. The changes are mostly small API additions or deprecations, but they tidy up the language and tooling.
Worth knowingstdlib
x/net/http2: deprecate WriteScheduler
- What changed
- The
x/net/http2.WriteSchedulerinterface and its implementations are marked deprecated. The round‑robin scheduler remains the default; the priority scheduler is buggy and will not be fixed. - Production impact
- The source does not say.
- Try it
- Run
go list -deps golang.org/x/net/http2and look forWriteSchedulerin the documentation. - Source
- github.com/golang/go/issues/67817
Explain it and run it
Understand it, then run it
The x/net/http2.WriteScheduler interface lets a server decide the order in which data is sent on HTTP/2 streams. Three built‑in schedulers exist: round‑robin (the default), random, and priority. The priority scheduler is buggy and the HTTP/2 prioritization it implements is no longer supported. Because of that, the Go project has marked the whole WriteScheduler mechanism as deprecated. The round‑robin scheduler will remain the default, and the priority scheduler will not be fixed.
Run it now
// This program demonstrates that the `x/net/http2.WriteScheduler` interface
// is still part of the external `golang.org/x/net/http2` package and is
// marked deprecated. Since the sandbox has no network access, we cannot
// import that package, so we simply print a message explaining the
// situation.
package main
import "fmt"
func main() {
fmt.Println("The WriteScheduler interface in golang.org/x/net/http2 is deprecated.")
}
What it printed when we ran it on Go 1.27.1
The WriteScheduler interface in golang.org/x/net/http2 is deprecated.
Run sends this program (for Solidity, the contract and its tests) to our own sandbox, where it is compiled and run once, with no network, and what it printed or the test report comes back here. Nothing is kept. Runs are counted per visitor for the day so everyone gets a turn; the details are on the legal page.
Worth knowingstdlib
net/url: URL.Clone, Values.Clone
- What changed
URL.Clone()returns a deep copy of anet/url.URL, copying theUserfield.Values.Clone()returns a deep copy of aurl.Valuesmap.- Production impact
- The source does not say.
- Try it
u2 := u1.Clone()and verifyu2 != u1and that modifications tou2do not affectu1.- Source
- github.com/golang/go/issues/73450
Explain it and run it
Understand it, then run it
URL.Clone() and Values.Clone() let you copy a net/url.URL or a url.Values without touching the original. Before, copying a URL was done by url2 := *url1 or by converting to a string and parsing again, both of which are confusing or expensive. Now you can simply call u.Clone() to get a new URL that is a deep copy, including the user info. The same applies to query parameters: v.Clone() copies a url.Values map so changes to the copy don’t affect the source.
Run it now
// This program demonstrates the new Clone methods for net/url.URL and url.Values.
// It runs in Go 1.27.1 and shows how to copy a URL and its query parameters safely.
package main
import (
"fmt"
"net/url"
)
func main() {
// Original URL with user info and query parameters.
orig, _ := url.Parse("https://user:[email protected]/path?foo=bar&baz=qux")
// Clone the URL. The returned value is a deep copy, including the User field.
cloned := orig.Clone()
// Modify the clone: change the path and add a query value.
cloned.Path = "/newpath"
cloned.Query().Set("new", "value")
// Show that the original URL is unchanged.
fmt.Println("Original:", orig.String())
fmt.Println("Cloned :", cloned.String())
// Clone the query values separately.
origVals := orig.Query()
clonedVals := origVals.Clone()
// Modify the cloned values.
clonedVals.Set("foo", "changed")
// The original values remain untouched.
fmt.Println("Original query:", origVals.Encode())
fmt.Println("Cloned query :", clonedVals.Encode())
}
What it printed when we ran it on Go 1.27.1
Original: https://user:[email protected]/path?foo=bar&baz=qux Cloned : https://user:[email protected]/newpath?foo=bar&baz=qux Original query: baz=qux&foo=bar Cloned query : baz=qux&foo=changed
Run sends this program (for Solidity, the contract and its tests) to our own sandbox, where it is compiled and run once, with no network, and what it printed or the test report comes back here. Nothing is kept. Runs are counted per visitor for the day so everyone gets a turn; the details are on the legal page.
Worth knowingtooling
cmd/vet: check for missing Err calls for bufio.Scanner and sql.Rows
- What changed
go vetnow warns whenbufio.Scanner.Err()orsql.Rows.Err()is not checked, except for known infallable readers.- Production impact
- The source does not say.
- Try it
- Write a small program that uses
bufio.Scannerwithout checkingErr()and rungo vet. - Source
- github.com/golang/go/issues/17747
Explain it and run it
Understand it, then run it
go vet now looks for places where you use a bufio.Scanner or sql.Rows and forget to call their Err() method. When you read from a scanner or iterate over rows, the Err() method tells you if an error happened during the operation. If you ignore that call, you might miss a failure and think the data was read correctly. The new check warns you when that call is missing, unless the underlying reader is known to never fail, such as a bytes.Buffer or strings.Reader.
Run it now
// This program demonstrates the new go vet check for missing Err calls
// on bufio.Scanner. It compiles and runs on Go 1.27.1. The program
// reads a string with a scanner, prints each token, and then checks
// scanner.Err() to ensure no scanning error occurred.
package main
import (
"bufio"
"fmt"
"strings"
)
func main() {
// Create a scanner over a simple string.
s := bufio.NewScanner(strings.NewReader("one two three"))
for s.Scan() { // use the scanner
fmt.Println(s.Text())
}
// Check the error after the loop.
if err := s.Err(); err != nil {
fmt.Println("scan error:", err)
}
}
What it printed when we ran it on Go 1.27.1
one two three
Run sends this program (for Solidity, the contract and its tests) to our own sandbox, where it is compiled and run once, with no network, and what it printed or the test report comes back here. Nothing is kept. Runs are counted per visitor for the day so everyone gets a turn; the details are on the legal page.
Nice to knowtooling
go/token: add (*File).String method
- What changed
token.File.String()now returns a brief description of the file, e.g.,"foo.go(1-0)".- Production impact
- The source does not say.
- Try it
fmt.Println(token.NewFileSet().AddFile("foo.go", -1, 0)).- Source
- github.com/golang/go/issues/76285
Explain it
Understand it, then run it
token.File.String() now prints a short description of a file instead of a long, unreadable struct. Previously, printing a *token.File produced a verbose output that included the file name, size, and internal slices, which was hard to read. The new method formats the file as "name(base-end)", for example "foo.go(1-0)". This makes debugging and logging easier for beginners who want to see which file a token.File refers to.
Nice to knowtooling
Using go fix to modernize Go code
- What changed
- Go 1.26 introduces a new implementation of
go fixthat can help modernize code. - Production impact
- The source does not say.
- Try it
- Run
go fix -ron a legacy Go file and observe the suggested changes. - Source
- go.dev/blog/gofix
Explain it and run it
Understand it, then run it
Run it now
// This program demonstrates the newexpr modernizer feature introduced in Go 1.26.
// It uses the new function with an expression argument and prints the result.
package main
import (
"fmt"
)
func main() {
// new(10) creates an int variable initialized to 10 and returns its address.
ptr := new(10)
fmt.Println(*ptr) // prints 10
}
What it printed when we ran it on Go 1.27.1
10
Run sends this program (for Solidity, the contract and its tests) to our own sandbox, where it is compiled and run once, with no network, and what it printed or the test report comes back here. Nothing is kept. Runs are counted per visitor for the day so everyone gets a turn; the details are on the legal page.
Exercise
Write a small program that parses a URL, clones it, modifies the clone, and prints both URLs to show that the original is unchanged.
package main
import (
"fmt"
"net/url"
)
func main() {
u, _ := url.Parse("https://example.com/path?foo=bar")
// TODO: clone u, change the clone, print both
}
Show a solution
package main
import (
"fmt"
"net/url"
)
func main() {
u, _ := url.Parse("https://example.com/path?foo=bar")
u2 := u.Clone()
u2.Path = "/newpath"
u2.RawQuery = "baz=qux"
fmt.Println("original:", u.String())
fmt.Println("clone :", u2.String())
}
What it printed when we ran it on Go 1.27.1
original: https://example.com/path?foo=bar clone : https://example.com/newpath?baz=qux
Run sends this program (for Solidity, the contract and its tests) to our own sandbox, where it is compiled and run once, with no network, and what it printed or the test report comes back here. Nothing is kept. Runs are counted per visitor for the day so everyone gets a turn; the details are on the legal page.
The 60-second version
Hello, everyone. This week the Go team made a handful of tidy updates. They deprecated the rarely used HTTP/2 write scheduler, added convenient clone methods for URLs and form values, gave file objects a cleaner string representation, and strengthened the vet tool to catch missing error checks on scanners and SQL rows. Finally, Go 1.26 ships a new go fix implementation that can help you modernize legacy code. These changes keep the language lean, improve safety, and make everyday tooling a bit more pleasant. Thank you.
Written by gpt-oss-20b · claims checked against the sources · archive, not individually reviewed