Improve limactl edit
#2593
Improve limactl edit
#2593
Conversation
| @@ -58,6 +59,8 @@ require ( | |||
| github.com/VividCortex/ewma v1.2.0 // indirect | |||
| github.com/a8m/envsubst v1.4.2 // indirect | |||
| github.com/alecthomas/participle/v2 v2.1.1 // indirect | |||
| github.com/bmatcuk/doublestar/v4 v4.6.0 // indirect | |||
| github.com/braydonk/yaml v0.7.0 // indirect | |||
There was a problem hiding this comment.
Not a huge fan of importing multiple YAML libraries
There was a problem hiding this comment.
Is this the fifth YAML library?
There was a problem hiding this comment.
Oh, is this the first time a YAML library forked from https://github.com/go-yaml/yaml is being added?
This comment was marked as off-topic.
This comment was marked as off-topic.
Sorry, something went wrong.
There was a problem hiding this comment.
I want to express my gratitude to Go for allowing the use of both a forked library and the original library simultaneously without causing issues. 👏🏻
There was a problem hiding this comment.
I think both of yqlib and yamlfmt brings enough value, to be allowed to bring-their-own-yaml to the party
|
Tests are failing on Windows because it is adding |
|
I like this new library, with some configuration it can actually keep the examples/default.yaml more or less as-is: .yamlfmt line_ending: lf
formatter:
indentless_arrays: true
retain_line_breaks: trueThe only change is some alignment on the comments, and I don't think that is too bad (and yamllint is OK too)
--- a/examples/default.yaml
+++ b/examples/default.yaml
@@ -264,10 +264,10 @@ containerd:
# Setting of instructions is supported like this: "qemu64,+ssse3".
# 🟢 Builtin default: hard-coded arch map with type (see the output of `limactl info | jq .defaultTemplate.cpuType`)
cpuType:
- # aarch64: "cortex-a72" # (or "host" when running on aarch64 host)
- # armv7l: "cortex-a7" # (or "host" when running on armv7l host)
- # riscv64: "rv64" # (or "host" when running on riscv64 host)
- # x86_64: "qemu64" # (or "host,-pdpe1gb" when running on x86_64 host)
+# aarch64: "cortex-a72" # (or "host" when running on aarch64 host)
+# armv7l: "cortex-a7" # (or "host" when running on armv7l host)
+# riscv64: "rv64" # (or "host" when running on riscv64 host)
+# x86_64: "qemu64" # (or "host,-pdpe1gb" when running on x86_64 host)
rosetta:
# Enable Rosetta for Linux (EXPERIMENTAL; will graduate from experimental in Lima v1.0).
@@ -456,8 +456,8 @@ hostResolver:
# predefined to specify the gateway address to the host.
# 🟢 Builtin default: null
hosts:
- # guest.name: 127.1.1.1
- # host.name: host.lima.internal
+ # guest.name: 127.1.1.1
+ # host.name: host.lima.internal
# If hostResolver.enabled is false, then the following rules apply for configuring dns:
# Explicitly set DNS addresses for qemu user-mode networking. By default qemu picks *one*
|
There was a problem hiding this comment.
I agree with @afbjorklund that the new libraries provide sufficient value to justify adding one more YAML library to the project.
I think the fact that limactl edit can now edit templates in addition to instances should be noted in the command descriptions:
- Use: "edit INSTANCE",
- Short: "Edit an instance of Lima",
+ Use: "edit INSTANCE|FILE.yaml",
+ Short: "Edit an instance of Lima or a template",
|
Please squash commits! |
Signed-off-by: Norio Nomura <[email protected]> `limactl edit`: support editing lima.yaml file This allows `limactl edit --set "..." <limayaml>` to be used instead of `yq -i eval "..." <limayaml>` in scripts like `hack/inject-cmdline-to-template.sh`. Signed-off-by: Norio Nomura <[email protected]> yqutil: Use `LF` for line endings regardless of the platform. Signed-off-by: Norio Nomura <[email protected]> `limactl edit`: update command description Signed-off-by: Norio Nomura <[email protected]> yqutil_test: fix typo Signed-off-by: Norio Nomura <[email protected]>
norio-nomura
force-pushed
the
improve-limactl-edit
branch
from
6cfba80
to
a10a8bb
Compare
|
@AkihiroSuda Do you have any objections to merging this? |
Super-hacky idea: Replace all empty lines with |
That's exactly the idea that |
Oh, it seems that the following hack works well: diff --git forkSrcPrefix/pkg/yqutil/yqutil.go forkDstPrefix/pkg/yqutil/yqutil.go
index 139d9fcbce3ff6c7ca3df38a04a1fb3ef36ccc1b..8787797f40b6e0a8fcaec1f06b203b25e8ba3c06 100644
--- forkSrcPrefix/pkg/yqutil/yqutil.go
+++ forkDstPrefix/pkg/yqutil/yqutil.go
@@ -1,6 +1,7 @@
package yqutil
import (
+ "bufio"
"bytes"
"fmt"
"os"
@@ -15,13 +16,17 @@ import (
// EvaluateExpression evaluates the yq expression, and returns the modified yaml.
func EvaluateExpression(expression string, content []byte) ([]byte, error) {
logrus.Debugf("Evaluating yq expression: %q", expression)
+ contentModified, err := replaceLineBreaksWithMagicString(content)
+ if err != nil {
+ return nil, err
+ }
tmpYAMLFile, err := os.CreateTemp("", "lima-yq-*.yaml")
if err != nil {
return nil, err
}
tmpYAMLPath := tmpYAMLFile.Name()
defer os.RemoveAll(tmpYAMLPath)
- _, err = tmpYAMLFile.Write(content)
+ _, err = tmpYAMLFile.Write(contentModified)
if err != nil {
tmpYAMLFile.Close()
return nil, err
@@ -94,3 +99,41 @@ func yamlfmt(content []byte) ([]byte, error) {
}
return formatter.Format(content)
}
+
+const yamlfmtLineBreakPlaceholder = "#magic___^_^___line"
+
+type paddinger struct {
+ strings.Builder
+}
+
+func (p *paddinger) adjust(txt string) {
+ var indentSize int
+ for i := 0; i < len(txt) && txt[i] == ' '; i++ { // yaml only allows space to indent.
+ indentSize++
+ }
+ // Grows if the given size is larger than us and always return the max padding.
+ for diff := indentSize - p.Len(); diff > 0; diff-- {
+ p.WriteByte(' ')
+ }
+}
+
+func replaceLineBreaksWithMagicString(content []byte) ([]byte, error) {
+ // hotfix: yq does not support line breaks in the middle of a string.
+ var buf bytes.Buffer
+ reader := bytes.NewReader(content)
+ scanner := bufio.NewScanner(reader)
+ var padding paddinger
+ for scanner.Scan() {
+ txt := scanner.Text()
+ padding.adjust(txt)
+ if strings.TrimSpace(txt) == "" { // line break or empty space line.
+ buf.WriteString(padding.String()) // prepend some padding incase literal multiline strings.
+ buf.WriteString(yamlfmtLineBreakPlaceholder)
+ buf.WriteString("\n")
+ } else {
+ buf.WriteString(txt)
+ buf.WriteString("\n")
+ }
+ }
+ return buf.Bytes(), scanner.Err()
+}
diff --git forkSrcPrefix/pkg/yqutil/yqutil_test.go forkDstPrefix/pkg/yqutil/yqutil_test.go
index fcf1260ec24519a847a9305d978144f0bbdbe7cc..287c51c6be2d13bd75a7984712f0d4aad492fb0b 100644
--- forkSrcPrefix/pkg/yqutil/yqutil_test.go
+++ forkDstPrefix/pkg/yqutil/yqutil_test.go
@@ -19,6 +19,7 @@ memory: null
expected := `
# CPUs
cpus: 2
+
# Memory size
memory: 2GiB
`🤔 (edited: add |
|
The changes after running diff --git forkSrcPrefix/examples/default.yaml forkDstPrefix/examples/default.yaml
index 5b269ff7e25530c16ce7a9c4fd00251a35850b0d..c46886072da9ff064754da989698e618daa787e8 100644
--- forkSrcPrefix/examples/default.yaml
+++ forkDstPrefix/examples/default.yaml
@@ -43,7 +43,7 @@ cpus: null
# Memory size
# 🟢 Builtin default: min("4GiB", half of host memory)
-memory: null
+memory: 8GiB
# Disk size
# 🟢 Builtin default: "100GiB"
@@ -264,10 +264,10 @@ containerd:
# Setting of instructions is supported like this: "qemu64,+ssse3".
# 🟢 Builtin default: hard-coded arch map with type (see the output of `limactl info | jq .defaultTemplate.cpuType`)
cpuType:
- # aarch64: "cortex-a72" # (or "host" when running on aarch64 host)
- # armv7l: "cortex-a7" # (or "host" when running on armv7l host)
- # riscv64: "rv64" # (or "host" when running on riscv64 host)
- # x86_64: "qemu64" # (or "host,-pdpe1gb" when running on x86_64 host)
+# aarch64: "cortex-a72" # (or "host" when running on aarch64 host)
+# armv7l: "cortex-a7" # (or "host" when running on armv7l host)
+# riscv64: "rv64" # (or "host" when running on riscv64 host)
+# x86_64: "qemu64" # (or "host,-pdpe1gb" when running on x86_64 host)
rosetta:
# Enable Rosetta for Linux (EXPERIMENTAL; will graduate from experimental in Lima v1.0).
@@ -456,8 +456,8 @@ hostResolver:
# predefined to specify the gateway address to the host.
# 🟢 Builtin default: null
hosts:
- # guest.name: 127.1.1.1
- # host.name: host.lima.internal
+ # guest.name: 127.1.1.1
+ # host.name: host.lima.internal
# If hostResolver.enabled is false, then the following rules apply for configuring dns:
# Explicitly set DNS addresses for qemu user-mode networking. By default qemu picks *one* |
If you create a PR with that hack (which I would like to see once the current PR is merged), then I would suggest that we should also fix up the templates so that the comments round-trip correctly (all comments should be in front of the setting they apply to, and not after it). Then we can add a check to CI to make sure they remain round-trippable. Something like find examples -name '*.yaml' -exec limactl edit --set 'del(.nothing)' {} \;
git diff-index --exit-code HEAD |
|
@AkihiroSuda I'll take the fact that you assigned this PR to the
So I wasn't sure if you still had objections to merging it or not. But I guess you are ok with it now. |
|
Thanks! 🙏🏻
Before doing that, I want to finalize #2532. |
Yes, ok to merge, sorry for delay |
limactl edit: support editinglima.yamlfileThis allows
limactl edit --set "..." <limayaml>to be used instead ofyq -i eval "..." <limayaml>in scripts likehack/inject-cmdline-to-template.sh.The changes made to
examples/ubuntu.yamlusinglimactl edit --setinhack/inject-cmdline-to-template.share as follows:The issue with line breaks disappearing cannot be fixed with yamlfmt, so it still remains. However, the sequence is no longer indented, and I personally think it has become usable.
I'm considering making a separate PR to change
hack/inject-cmdline-to-template.shfromyq -i evaltolimactl edit --setafter #2562, which is likely to have overlapping changes, is resolved.