platform_build_soong/ui/status/ninja.go

275 lines
7.3 KiB
Go
Raw Normal View History

Add a unified status reporting UI This adds a new status package that merges the running of "actions" (ninja calls them edges) of multiple tools into one view of the current state, and gives that to a number of different outputs. For inputs: Kati's output parser has been rewritten (and moved) to map onto the StartAction/FinishAction API. A byproduct of this is that the build servers should be able to extract errors from Kati better, since they look like the errors that Ninja used to write. Ninja is no longer directly connected to the terminal, but its output is read via the protobuf frontend API, so it's just another tool whose output becomes merged together. multiproduct_kati loses its custom status routines, and uses the common one instead. For outputs: The primary output is the ui/terminal.Status type, which along with ui/terminal.Writer now controls everything about the terminal output. Today, this doesn't really change any behaviors, but having all terminal output going through here allows a more complicated (multi-line / full window) status display in the future. The tracer acts as an output of the status package, tracing all the action start / finish events. This replaces reading the .ninja_log file, so it now properly handles multiple output files from a single action. A new rotated log file (out/error.log, or out/dist/logs/error.log) just contains a description of all of the errors that happened during the current build. Another new compressed and rotated log file (out/verbose.log.gz, or out/dist/logs/verbose.log.gz) contains the full verbose (showcommands) log of every execution run by the build. Since this is now written on every build, the showcommands argument is now ignored -- if you want to get the commands run, look at the log file after the build. Test: m Test: <built-in tests> Test: NINJA_ARGS="-t list" m Test: check the build.trace.gz Test: check the new log files Change-Id: If1d8994890d43ef68f65aa10ddd8e6e06dc7013a
2018-05-18 01:37:09 +02:00
// Copyright 2018 Google Inc. All rights reserved.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
package status
import (
"bufio"
"fmt"
"io"
"os"
"regexp"
"strings"
Add a unified status reporting UI This adds a new status package that merges the running of "actions" (ninja calls them edges) of multiple tools into one view of the current state, and gives that to a number of different outputs. For inputs: Kati's output parser has been rewritten (and moved) to map onto the StartAction/FinishAction API. A byproduct of this is that the build servers should be able to extract errors from Kati better, since they look like the errors that Ninja used to write. Ninja is no longer directly connected to the terminal, but its output is read via the protobuf frontend API, so it's just another tool whose output becomes merged together. multiproduct_kati loses its custom status routines, and uses the common one instead. For outputs: The primary output is the ui/terminal.Status type, which along with ui/terminal.Writer now controls everything about the terminal output. Today, this doesn't really change any behaviors, but having all terminal output going through here allows a more complicated (multi-line / full window) status display in the future. The tracer acts as an output of the status package, tracing all the action start / finish events. This replaces reading the .ninja_log file, so it now properly handles multiple output files from a single action. A new rotated log file (out/error.log, or out/dist/logs/error.log) just contains a description of all of the errors that happened during the current build. Another new compressed and rotated log file (out/verbose.log.gz, or out/dist/logs/verbose.log.gz) contains the full verbose (showcommands) log of every execution run by the build. Since this is now written on every build, the showcommands argument is now ignored -- if you want to get the commands run, look at the log file after the build. Test: m Test: <built-in tests> Test: NINJA_ARGS="-t list" m Test: check the build.trace.gz Test: check the new log files Change-Id: If1d8994890d43ef68f65aa10ddd8e6e06dc7013a
2018-05-18 01:37:09 +02:00
"syscall"
"time"
Add a unified status reporting UI This adds a new status package that merges the running of "actions" (ninja calls them edges) of multiple tools into one view of the current state, and gives that to a number of different outputs. For inputs: Kati's output parser has been rewritten (and moved) to map onto the StartAction/FinishAction API. A byproduct of this is that the build servers should be able to extract errors from Kati better, since they look like the errors that Ninja used to write. Ninja is no longer directly connected to the terminal, but its output is read via the protobuf frontend API, so it's just another tool whose output becomes merged together. multiproduct_kati loses its custom status routines, and uses the common one instead. For outputs: The primary output is the ui/terminal.Status type, which along with ui/terminal.Writer now controls everything about the terminal output. Today, this doesn't really change any behaviors, but having all terminal output going through here allows a more complicated (multi-line / full window) status display in the future. The tracer acts as an output of the status package, tracing all the action start / finish events. This replaces reading the .ninja_log file, so it now properly handles multiple output files from a single action. A new rotated log file (out/error.log, or out/dist/logs/error.log) just contains a description of all of the errors that happened during the current build. Another new compressed and rotated log file (out/verbose.log.gz, or out/dist/logs/verbose.log.gz) contains the full verbose (showcommands) log of every execution run by the build. Since this is now written on every build, the showcommands argument is now ignored -- if you want to get the commands run, look at the log file after the build. Test: m Test: <built-in tests> Test: NINJA_ARGS="-t list" m Test: check the build.trace.gz Test: check the new log files Change-Id: If1d8994890d43ef68f65aa10ddd8e6e06dc7013a
2018-05-18 01:37:09 +02:00
"github.com/golang/protobuf/proto"
"android/soong/ui/logger"
"android/soong/ui/status/ninja_frontend"
)
// NewNinjaReader reads the protobuf frontend format from ninja and translates it
Add a unified status reporting UI This adds a new status package that merges the running of "actions" (ninja calls them edges) of multiple tools into one view of the current state, and gives that to a number of different outputs. For inputs: Kati's output parser has been rewritten (and moved) to map onto the StartAction/FinishAction API. A byproduct of this is that the build servers should be able to extract errors from Kati better, since they look like the errors that Ninja used to write. Ninja is no longer directly connected to the terminal, but its output is read via the protobuf frontend API, so it's just another tool whose output becomes merged together. multiproduct_kati loses its custom status routines, and uses the common one instead. For outputs: The primary output is the ui/terminal.Status type, which along with ui/terminal.Writer now controls everything about the terminal output. Today, this doesn't really change any behaviors, but having all terminal output going through here allows a more complicated (multi-line / full window) status display in the future. The tracer acts as an output of the status package, tracing all the action start / finish events. This replaces reading the .ninja_log file, so it now properly handles multiple output files from a single action. A new rotated log file (out/error.log, or out/dist/logs/error.log) just contains a description of all of the errors that happened during the current build. Another new compressed and rotated log file (out/verbose.log.gz, or out/dist/logs/verbose.log.gz) contains the full verbose (showcommands) log of every execution run by the build. Since this is now written on every build, the showcommands argument is now ignored -- if you want to get the commands run, look at the log file after the build. Test: m Test: <built-in tests> Test: NINJA_ARGS="-t list" m Test: check the build.trace.gz Test: check the new log files Change-Id: If1d8994890d43ef68f65aa10ddd8e6e06dc7013a
2018-05-18 01:37:09 +02:00
// into calls on the ToolStatus API.
func NewNinjaReader(ctx logger.Logger, status ToolStatus, fifo string) *NinjaReader {
Add a unified status reporting UI This adds a new status package that merges the running of "actions" (ninja calls them edges) of multiple tools into one view of the current state, and gives that to a number of different outputs. For inputs: Kati's output parser has been rewritten (and moved) to map onto the StartAction/FinishAction API. A byproduct of this is that the build servers should be able to extract errors from Kati better, since they look like the errors that Ninja used to write. Ninja is no longer directly connected to the terminal, but its output is read via the protobuf frontend API, so it's just another tool whose output becomes merged together. multiproduct_kati loses its custom status routines, and uses the common one instead. For outputs: The primary output is the ui/terminal.Status type, which along with ui/terminal.Writer now controls everything about the terminal output. Today, this doesn't really change any behaviors, but having all terminal output going through here allows a more complicated (multi-line / full window) status display in the future. The tracer acts as an output of the status package, tracing all the action start / finish events. This replaces reading the .ninja_log file, so it now properly handles multiple output files from a single action. A new rotated log file (out/error.log, or out/dist/logs/error.log) just contains a description of all of the errors that happened during the current build. Another new compressed and rotated log file (out/verbose.log.gz, or out/dist/logs/verbose.log.gz) contains the full verbose (showcommands) log of every execution run by the build. Since this is now written on every build, the showcommands argument is now ignored -- if you want to get the commands run, look at the log file after the build. Test: m Test: <built-in tests> Test: NINJA_ARGS="-t list" m Test: check the build.trace.gz Test: check the new log files Change-Id: If1d8994890d43ef68f65aa10ddd8e6e06dc7013a
2018-05-18 01:37:09 +02:00
os.Remove(fifo)
err := syscall.Mkfifo(fifo, 0666)
if err != nil {
ctx.Fatalf("Failed to mkfifo(%q): %v", fifo, err)
}
n := &NinjaReader{
status: status,
fifo: fifo,
done: make(chan bool),
cancel: make(chan bool),
}
go n.run()
return n
Add a unified status reporting UI This adds a new status package that merges the running of "actions" (ninja calls them edges) of multiple tools into one view of the current state, and gives that to a number of different outputs. For inputs: Kati's output parser has been rewritten (and moved) to map onto the StartAction/FinishAction API. A byproduct of this is that the build servers should be able to extract errors from Kati better, since they look like the errors that Ninja used to write. Ninja is no longer directly connected to the terminal, but its output is read via the protobuf frontend API, so it's just another tool whose output becomes merged together. multiproduct_kati loses its custom status routines, and uses the common one instead. For outputs: The primary output is the ui/terminal.Status type, which along with ui/terminal.Writer now controls everything about the terminal output. Today, this doesn't really change any behaviors, but having all terminal output going through here allows a more complicated (multi-line / full window) status display in the future. The tracer acts as an output of the status package, tracing all the action start / finish events. This replaces reading the .ninja_log file, so it now properly handles multiple output files from a single action. A new rotated log file (out/error.log, or out/dist/logs/error.log) just contains a description of all of the errors that happened during the current build. Another new compressed and rotated log file (out/verbose.log.gz, or out/dist/logs/verbose.log.gz) contains the full verbose (showcommands) log of every execution run by the build. Since this is now written on every build, the showcommands argument is now ignored -- if you want to get the commands run, look at the log file after the build. Test: m Test: <built-in tests> Test: NINJA_ARGS="-t list" m Test: check the build.trace.gz Test: check the new log files Change-Id: If1d8994890d43ef68f65aa10ddd8e6e06dc7013a
2018-05-18 01:37:09 +02:00
}
type NinjaReader struct {
status ToolStatus
fifo string
done chan bool
cancel chan bool
}
const NINJA_READER_CLOSE_TIMEOUT = 5 * time.Second
// Close waits for NinjaReader to finish reading from the fifo, or 5 seconds.
func (n *NinjaReader) Close() {
// Signal the goroutine to stop if it is blocking opening the fifo.
close(n.cancel)
timeoutCh := time.After(NINJA_READER_CLOSE_TIMEOUT)
select {
case <-n.done:
// Nothing
case <-timeoutCh:
n.status.Error(fmt.Sprintf("ninja fifo didn't finish after %s", NINJA_READER_CLOSE_TIMEOUT.String()))
}
return
}
func (n *NinjaReader) run() {
defer close(n.done)
// Opening the fifo can block forever if ninja never opens the write end, do it in a goroutine so this
// method can exit on cancel.
fileCh := make(chan *os.File)
go func() {
f, err := os.Open(n.fifo)
if err != nil {
n.status.Error(fmt.Sprintf("Failed to open fifo: %v", err))
close(fileCh)
return
}
fileCh <- f
}()
var f *os.File
select {
case f = <-fileCh:
// Nothing
case <-n.cancel:
return
Add a unified status reporting UI This adds a new status package that merges the running of "actions" (ninja calls them edges) of multiple tools into one view of the current state, and gives that to a number of different outputs. For inputs: Kati's output parser has been rewritten (and moved) to map onto the StartAction/FinishAction API. A byproduct of this is that the build servers should be able to extract errors from Kati better, since they look like the errors that Ninja used to write. Ninja is no longer directly connected to the terminal, but its output is read via the protobuf frontend API, so it's just another tool whose output becomes merged together. multiproduct_kati loses its custom status routines, and uses the common one instead. For outputs: The primary output is the ui/terminal.Status type, which along with ui/terminal.Writer now controls everything about the terminal output. Today, this doesn't really change any behaviors, but having all terminal output going through here allows a more complicated (multi-line / full window) status display in the future. The tracer acts as an output of the status package, tracing all the action start / finish events. This replaces reading the .ninja_log file, so it now properly handles multiple output files from a single action. A new rotated log file (out/error.log, or out/dist/logs/error.log) just contains a description of all of the errors that happened during the current build. Another new compressed and rotated log file (out/verbose.log.gz, or out/dist/logs/verbose.log.gz) contains the full verbose (showcommands) log of every execution run by the build. Since this is now written on every build, the showcommands argument is now ignored -- if you want to get the commands run, look at the log file after the build. Test: m Test: <built-in tests> Test: NINJA_ARGS="-t list" m Test: check the build.trace.gz Test: check the new log files Change-Id: If1d8994890d43ef68f65aa10ddd8e6e06dc7013a
2018-05-18 01:37:09 +02:00
}
Add a unified status reporting UI This adds a new status package that merges the running of "actions" (ninja calls them edges) of multiple tools into one view of the current state, and gives that to a number of different outputs. For inputs: Kati's output parser has been rewritten (and moved) to map onto the StartAction/FinishAction API. A byproduct of this is that the build servers should be able to extract errors from Kati better, since they look like the errors that Ninja used to write. Ninja is no longer directly connected to the terminal, but its output is read via the protobuf frontend API, so it's just another tool whose output becomes merged together. multiproduct_kati loses its custom status routines, and uses the common one instead. For outputs: The primary output is the ui/terminal.Status type, which along with ui/terminal.Writer now controls everything about the terminal output. Today, this doesn't really change any behaviors, but having all terminal output going through here allows a more complicated (multi-line / full window) status display in the future. The tracer acts as an output of the status package, tracing all the action start / finish events. This replaces reading the .ninja_log file, so it now properly handles multiple output files from a single action. A new rotated log file (out/error.log, or out/dist/logs/error.log) just contains a description of all of the errors that happened during the current build. Another new compressed and rotated log file (out/verbose.log.gz, or out/dist/logs/verbose.log.gz) contains the full verbose (showcommands) log of every execution run by the build. Since this is now written on every build, the showcommands argument is now ignored -- if you want to get the commands run, look at the log file after the build. Test: m Test: <built-in tests> Test: NINJA_ARGS="-t list" m Test: check the build.trace.gz Test: check the new log files Change-Id: If1d8994890d43ef68f65aa10ddd8e6e06dc7013a
2018-05-18 01:37:09 +02:00
defer f.Close()
r := bufio.NewReader(f)
running := map[uint32]*Action{}
for {
size, err := readVarInt(r)
if err != nil {
if err != io.EOF {
n.status.Error(fmt.Sprintf("Got error reading from ninja: %s", err))
Add a unified status reporting UI This adds a new status package that merges the running of "actions" (ninja calls them edges) of multiple tools into one view of the current state, and gives that to a number of different outputs. For inputs: Kati's output parser has been rewritten (and moved) to map onto the StartAction/FinishAction API. A byproduct of this is that the build servers should be able to extract errors from Kati better, since they look like the errors that Ninja used to write. Ninja is no longer directly connected to the terminal, but its output is read via the protobuf frontend API, so it's just another tool whose output becomes merged together. multiproduct_kati loses its custom status routines, and uses the common one instead. For outputs: The primary output is the ui/terminal.Status type, which along with ui/terminal.Writer now controls everything about the terminal output. Today, this doesn't really change any behaviors, but having all terminal output going through here allows a more complicated (multi-line / full window) status display in the future. The tracer acts as an output of the status package, tracing all the action start / finish events. This replaces reading the .ninja_log file, so it now properly handles multiple output files from a single action. A new rotated log file (out/error.log, or out/dist/logs/error.log) just contains a description of all of the errors that happened during the current build. Another new compressed and rotated log file (out/verbose.log.gz, or out/dist/logs/verbose.log.gz) contains the full verbose (showcommands) log of every execution run by the build. Since this is now written on every build, the showcommands argument is now ignored -- if you want to get the commands run, look at the log file after the build. Test: m Test: <built-in tests> Test: NINJA_ARGS="-t list" m Test: check the build.trace.gz Test: check the new log files Change-Id: If1d8994890d43ef68f65aa10ddd8e6e06dc7013a
2018-05-18 01:37:09 +02:00
}
return
}
buf := make([]byte, size)
_, err = io.ReadFull(r, buf)
if err != nil {
if err == io.EOF {
n.status.Print(fmt.Sprintf("Missing message of size %d from ninja\n", size))
Add a unified status reporting UI This adds a new status package that merges the running of "actions" (ninja calls them edges) of multiple tools into one view of the current state, and gives that to a number of different outputs. For inputs: Kati's output parser has been rewritten (and moved) to map onto the StartAction/FinishAction API. A byproduct of this is that the build servers should be able to extract errors from Kati better, since they look like the errors that Ninja used to write. Ninja is no longer directly connected to the terminal, but its output is read via the protobuf frontend API, so it's just another tool whose output becomes merged together. multiproduct_kati loses its custom status routines, and uses the common one instead. For outputs: The primary output is the ui/terminal.Status type, which along with ui/terminal.Writer now controls everything about the terminal output. Today, this doesn't really change any behaviors, but having all terminal output going through here allows a more complicated (multi-line / full window) status display in the future. The tracer acts as an output of the status package, tracing all the action start / finish events. This replaces reading the .ninja_log file, so it now properly handles multiple output files from a single action. A new rotated log file (out/error.log, or out/dist/logs/error.log) just contains a description of all of the errors that happened during the current build. Another new compressed and rotated log file (out/verbose.log.gz, or out/dist/logs/verbose.log.gz) contains the full verbose (showcommands) log of every execution run by the build. Since this is now written on every build, the showcommands argument is now ignored -- if you want to get the commands run, look at the log file after the build. Test: m Test: <built-in tests> Test: NINJA_ARGS="-t list" m Test: check the build.trace.gz Test: check the new log files Change-Id: If1d8994890d43ef68f65aa10ddd8e6e06dc7013a
2018-05-18 01:37:09 +02:00
} else {
n.status.Error(fmt.Sprintf("Got error reading from ninja: %s", err))
Add a unified status reporting UI This adds a new status package that merges the running of "actions" (ninja calls them edges) of multiple tools into one view of the current state, and gives that to a number of different outputs. For inputs: Kati's output parser has been rewritten (and moved) to map onto the StartAction/FinishAction API. A byproduct of this is that the build servers should be able to extract errors from Kati better, since they look like the errors that Ninja used to write. Ninja is no longer directly connected to the terminal, but its output is read via the protobuf frontend API, so it's just another tool whose output becomes merged together. multiproduct_kati loses its custom status routines, and uses the common one instead. For outputs: The primary output is the ui/terminal.Status type, which along with ui/terminal.Writer now controls everything about the terminal output. Today, this doesn't really change any behaviors, but having all terminal output going through here allows a more complicated (multi-line / full window) status display in the future. The tracer acts as an output of the status package, tracing all the action start / finish events. This replaces reading the .ninja_log file, so it now properly handles multiple output files from a single action. A new rotated log file (out/error.log, or out/dist/logs/error.log) just contains a description of all of the errors that happened during the current build. Another new compressed and rotated log file (out/verbose.log.gz, or out/dist/logs/verbose.log.gz) contains the full verbose (showcommands) log of every execution run by the build. Since this is now written on every build, the showcommands argument is now ignored -- if you want to get the commands run, look at the log file after the build. Test: m Test: <built-in tests> Test: NINJA_ARGS="-t list" m Test: check the build.trace.gz Test: check the new log files Change-Id: If1d8994890d43ef68f65aa10ddd8e6e06dc7013a
2018-05-18 01:37:09 +02:00
}
return
}
msg := &ninja_frontend.Status{}
err = proto.Unmarshal(buf, msg)
if err != nil {
n.status.Print(fmt.Sprintf("Error reading message from ninja: %v", err))
Add a unified status reporting UI This adds a new status package that merges the running of "actions" (ninja calls them edges) of multiple tools into one view of the current state, and gives that to a number of different outputs. For inputs: Kati's output parser has been rewritten (and moved) to map onto the StartAction/FinishAction API. A byproduct of this is that the build servers should be able to extract errors from Kati better, since they look like the errors that Ninja used to write. Ninja is no longer directly connected to the terminal, but its output is read via the protobuf frontend API, so it's just another tool whose output becomes merged together. multiproduct_kati loses its custom status routines, and uses the common one instead. For outputs: The primary output is the ui/terminal.Status type, which along with ui/terminal.Writer now controls everything about the terminal output. Today, this doesn't really change any behaviors, but having all terminal output going through here allows a more complicated (multi-line / full window) status display in the future. The tracer acts as an output of the status package, tracing all the action start / finish events. This replaces reading the .ninja_log file, so it now properly handles multiple output files from a single action. A new rotated log file (out/error.log, or out/dist/logs/error.log) just contains a description of all of the errors that happened during the current build. Another new compressed and rotated log file (out/verbose.log.gz, or out/dist/logs/verbose.log.gz) contains the full verbose (showcommands) log of every execution run by the build. Since this is now written on every build, the showcommands argument is now ignored -- if you want to get the commands run, look at the log file after the build. Test: m Test: <built-in tests> Test: NINJA_ARGS="-t list" m Test: check the build.trace.gz Test: check the new log files Change-Id: If1d8994890d43ef68f65aa10ddd8e6e06dc7013a
2018-05-18 01:37:09 +02:00
continue
}
// Ignore msg.BuildStarted
if msg.TotalEdges != nil {
n.status.SetTotalActions(int(msg.TotalEdges.GetTotalEdges()))
Add a unified status reporting UI This adds a new status package that merges the running of "actions" (ninja calls them edges) of multiple tools into one view of the current state, and gives that to a number of different outputs. For inputs: Kati's output parser has been rewritten (and moved) to map onto the StartAction/FinishAction API. A byproduct of this is that the build servers should be able to extract errors from Kati better, since they look like the errors that Ninja used to write. Ninja is no longer directly connected to the terminal, but its output is read via the protobuf frontend API, so it's just another tool whose output becomes merged together. multiproduct_kati loses its custom status routines, and uses the common one instead. For outputs: The primary output is the ui/terminal.Status type, which along with ui/terminal.Writer now controls everything about the terminal output. Today, this doesn't really change any behaviors, but having all terminal output going through here allows a more complicated (multi-line / full window) status display in the future. The tracer acts as an output of the status package, tracing all the action start / finish events. This replaces reading the .ninja_log file, so it now properly handles multiple output files from a single action. A new rotated log file (out/error.log, or out/dist/logs/error.log) just contains a description of all of the errors that happened during the current build. Another new compressed and rotated log file (out/verbose.log.gz, or out/dist/logs/verbose.log.gz) contains the full verbose (showcommands) log of every execution run by the build. Since this is now written on every build, the showcommands argument is now ignored -- if you want to get the commands run, look at the log file after the build. Test: m Test: <built-in tests> Test: NINJA_ARGS="-t list" m Test: check the build.trace.gz Test: check the new log files Change-Id: If1d8994890d43ef68f65aa10ddd8e6e06dc7013a
2018-05-18 01:37:09 +02:00
}
if msg.EdgeStarted != nil {
action := &Action{
Description: msg.EdgeStarted.GetDesc(),
Outputs: msg.EdgeStarted.Outputs,
Inputs: msg.EdgeStarted.Inputs,
Add a unified status reporting UI This adds a new status package that merges the running of "actions" (ninja calls them edges) of multiple tools into one view of the current state, and gives that to a number of different outputs. For inputs: Kati's output parser has been rewritten (and moved) to map onto the StartAction/FinishAction API. A byproduct of this is that the build servers should be able to extract errors from Kati better, since they look like the errors that Ninja used to write. Ninja is no longer directly connected to the terminal, but its output is read via the protobuf frontend API, so it's just another tool whose output becomes merged together. multiproduct_kati loses its custom status routines, and uses the common one instead. For outputs: The primary output is the ui/terminal.Status type, which along with ui/terminal.Writer now controls everything about the terminal output. Today, this doesn't really change any behaviors, but having all terminal output going through here allows a more complicated (multi-line / full window) status display in the future. The tracer acts as an output of the status package, tracing all the action start / finish events. This replaces reading the .ninja_log file, so it now properly handles multiple output files from a single action. A new rotated log file (out/error.log, or out/dist/logs/error.log) just contains a description of all of the errors that happened during the current build. Another new compressed and rotated log file (out/verbose.log.gz, or out/dist/logs/verbose.log.gz) contains the full verbose (showcommands) log of every execution run by the build. Since this is now written on every build, the showcommands argument is now ignored -- if you want to get the commands run, look at the log file after the build. Test: m Test: <built-in tests> Test: NINJA_ARGS="-t list" m Test: check the build.trace.gz Test: check the new log files Change-Id: If1d8994890d43ef68f65aa10ddd8e6e06dc7013a
2018-05-18 01:37:09 +02:00
Command: msg.EdgeStarted.GetCommand(),
}
n.status.StartAction(action)
Add a unified status reporting UI This adds a new status package that merges the running of "actions" (ninja calls them edges) of multiple tools into one view of the current state, and gives that to a number of different outputs. For inputs: Kati's output parser has been rewritten (and moved) to map onto the StartAction/FinishAction API. A byproduct of this is that the build servers should be able to extract errors from Kati better, since they look like the errors that Ninja used to write. Ninja is no longer directly connected to the terminal, but its output is read via the protobuf frontend API, so it's just another tool whose output becomes merged together. multiproduct_kati loses its custom status routines, and uses the common one instead. For outputs: The primary output is the ui/terminal.Status type, which along with ui/terminal.Writer now controls everything about the terminal output. Today, this doesn't really change any behaviors, but having all terminal output going through here allows a more complicated (multi-line / full window) status display in the future. The tracer acts as an output of the status package, tracing all the action start / finish events. This replaces reading the .ninja_log file, so it now properly handles multiple output files from a single action. A new rotated log file (out/error.log, or out/dist/logs/error.log) just contains a description of all of the errors that happened during the current build. Another new compressed and rotated log file (out/verbose.log.gz, or out/dist/logs/verbose.log.gz) contains the full verbose (showcommands) log of every execution run by the build. Since this is now written on every build, the showcommands argument is now ignored -- if you want to get the commands run, look at the log file after the build. Test: m Test: <built-in tests> Test: NINJA_ARGS="-t list" m Test: check the build.trace.gz Test: check the new log files Change-Id: If1d8994890d43ef68f65aa10ddd8e6e06dc7013a
2018-05-18 01:37:09 +02:00
running[msg.EdgeStarted.GetId()] = action
}
if msg.EdgeFinished != nil {
if started, ok := running[msg.EdgeFinished.GetId()]; ok {
delete(running, msg.EdgeFinished.GetId())
var err error
exitCode := int(msg.EdgeFinished.GetStatus())
if exitCode != 0 {
err = fmt.Errorf("exited with code: %d", exitCode)
}
outputWithErrorHint := errorHintGenerator.GetOutputWithErrorHint(msg.EdgeFinished.GetOutput(), exitCode)
n.status.FinishAction(ActionResult{
Add a unified status reporting UI This adds a new status package that merges the running of "actions" (ninja calls them edges) of multiple tools into one view of the current state, and gives that to a number of different outputs. For inputs: Kati's output parser has been rewritten (and moved) to map onto the StartAction/FinishAction API. A byproduct of this is that the build servers should be able to extract errors from Kati better, since they look like the errors that Ninja used to write. Ninja is no longer directly connected to the terminal, but its output is read via the protobuf frontend API, so it's just another tool whose output becomes merged together. multiproduct_kati loses its custom status routines, and uses the common one instead. For outputs: The primary output is the ui/terminal.Status type, which along with ui/terminal.Writer now controls everything about the terminal output. Today, this doesn't really change any behaviors, but having all terminal output going through here allows a more complicated (multi-line / full window) status display in the future. The tracer acts as an output of the status package, tracing all the action start / finish events. This replaces reading the .ninja_log file, so it now properly handles multiple output files from a single action. A new rotated log file (out/error.log, or out/dist/logs/error.log) just contains a description of all of the errors that happened during the current build. Another new compressed and rotated log file (out/verbose.log.gz, or out/dist/logs/verbose.log.gz) contains the full verbose (showcommands) log of every execution run by the build. Since this is now written on every build, the showcommands argument is now ignored -- if you want to get the commands run, look at the log file after the build. Test: m Test: <built-in tests> Test: NINJA_ARGS="-t list" m Test: check the build.trace.gz Test: check the new log files Change-Id: If1d8994890d43ef68f65aa10ddd8e6e06dc7013a
2018-05-18 01:37:09 +02:00
Action: started,
Output: outputWithErrorHint,
Add a unified status reporting UI This adds a new status package that merges the running of "actions" (ninja calls them edges) of multiple tools into one view of the current state, and gives that to a number of different outputs. For inputs: Kati's output parser has been rewritten (and moved) to map onto the StartAction/FinishAction API. A byproduct of this is that the build servers should be able to extract errors from Kati better, since they look like the errors that Ninja used to write. Ninja is no longer directly connected to the terminal, but its output is read via the protobuf frontend API, so it's just another tool whose output becomes merged together. multiproduct_kati loses its custom status routines, and uses the common one instead. For outputs: The primary output is the ui/terminal.Status type, which along with ui/terminal.Writer now controls everything about the terminal output. Today, this doesn't really change any behaviors, but having all terminal output going through here allows a more complicated (multi-line / full window) status display in the future. The tracer acts as an output of the status package, tracing all the action start / finish events. This replaces reading the .ninja_log file, so it now properly handles multiple output files from a single action. A new rotated log file (out/error.log, or out/dist/logs/error.log) just contains a description of all of the errors that happened during the current build. Another new compressed and rotated log file (out/verbose.log.gz, or out/dist/logs/verbose.log.gz) contains the full verbose (showcommands) log of every execution run by the build. Since this is now written on every build, the showcommands argument is now ignored -- if you want to get the commands run, look at the log file after the build. Test: m Test: <built-in tests> Test: NINJA_ARGS="-t list" m Test: check the build.trace.gz Test: check the new log files Change-Id: If1d8994890d43ef68f65aa10ddd8e6e06dc7013a
2018-05-18 01:37:09 +02:00
Error: err,
Stats: ActionResultStats{
UserTime: msg.EdgeFinished.GetUserTime(),
SystemTime: msg.EdgeFinished.GetSystemTime(),
MaxRssKB: msg.EdgeFinished.GetMaxRssKb(),
MinorPageFaults: msg.EdgeFinished.GetMinorPageFaults(),
MajorPageFaults: msg.EdgeFinished.GetMajorPageFaults(),
IOInputKB: msg.EdgeFinished.GetIoInputKb(),
IOOutputKB: msg.EdgeFinished.GetIoOutputKb(),
VoluntaryContextSwitches: msg.EdgeFinished.GetVoluntaryContextSwitches(),
InvoluntaryContextSwitches: msg.EdgeFinished.GetInvoluntaryContextSwitches(),
},
Add a unified status reporting UI This adds a new status package that merges the running of "actions" (ninja calls them edges) of multiple tools into one view of the current state, and gives that to a number of different outputs. For inputs: Kati's output parser has been rewritten (and moved) to map onto the StartAction/FinishAction API. A byproduct of this is that the build servers should be able to extract errors from Kati better, since they look like the errors that Ninja used to write. Ninja is no longer directly connected to the terminal, but its output is read via the protobuf frontend API, so it's just another tool whose output becomes merged together. multiproduct_kati loses its custom status routines, and uses the common one instead. For outputs: The primary output is the ui/terminal.Status type, which along with ui/terminal.Writer now controls everything about the terminal output. Today, this doesn't really change any behaviors, but having all terminal output going through here allows a more complicated (multi-line / full window) status display in the future. The tracer acts as an output of the status package, tracing all the action start / finish events. This replaces reading the .ninja_log file, so it now properly handles multiple output files from a single action. A new rotated log file (out/error.log, or out/dist/logs/error.log) just contains a description of all of the errors that happened during the current build. Another new compressed and rotated log file (out/verbose.log.gz, or out/dist/logs/verbose.log.gz) contains the full verbose (showcommands) log of every execution run by the build. Since this is now written on every build, the showcommands argument is now ignored -- if you want to get the commands run, look at the log file after the build. Test: m Test: <built-in tests> Test: NINJA_ARGS="-t list" m Test: check the build.trace.gz Test: check the new log files Change-Id: If1d8994890d43ef68f65aa10ddd8e6e06dc7013a
2018-05-18 01:37:09 +02:00
})
}
}
if msg.Message != nil {
message := "ninja: " + msg.Message.GetMessage()
switch msg.Message.GetLevel() {
case ninja_frontend.Status_Message_INFO:
n.status.Status(message)
Add a unified status reporting UI This adds a new status package that merges the running of "actions" (ninja calls them edges) of multiple tools into one view of the current state, and gives that to a number of different outputs. For inputs: Kati's output parser has been rewritten (and moved) to map onto the StartAction/FinishAction API. A byproduct of this is that the build servers should be able to extract errors from Kati better, since they look like the errors that Ninja used to write. Ninja is no longer directly connected to the terminal, but its output is read via the protobuf frontend API, so it's just another tool whose output becomes merged together. multiproduct_kati loses its custom status routines, and uses the common one instead. For outputs: The primary output is the ui/terminal.Status type, which along with ui/terminal.Writer now controls everything about the terminal output. Today, this doesn't really change any behaviors, but having all terminal output going through here allows a more complicated (multi-line / full window) status display in the future. The tracer acts as an output of the status package, tracing all the action start / finish events. This replaces reading the .ninja_log file, so it now properly handles multiple output files from a single action. A new rotated log file (out/error.log, or out/dist/logs/error.log) just contains a description of all of the errors that happened during the current build. Another new compressed and rotated log file (out/verbose.log.gz, or out/dist/logs/verbose.log.gz) contains the full verbose (showcommands) log of every execution run by the build. Since this is now written on every build, the showcommands argument is now ignored -- if you want to get the commands run, look at the log file after the build. Test: m Test: <built-in tests> Test: NINJA_ARGS="-t list" m Test: check the build.trace.gz Test: check the new log files Change-Id: If1d8994890d43ef68f65aa10ddd8e6e06dc7013a
2018-05-18 01:37:09 +02:00
case ninja_frontend.Status_Message_WARNING:
n.status.Print("warning: " + message)
Add a unified status reporting UI This adds a new status package that merges the running of "actions" (ninja calls them edges) of multiple tools into one view of the current state, and gives that to a number of different outputs. For inputs: Kati's output parser has been rewritten (and moved) to map onto the StartAction/FinishAction API. A byproduct of this is that the build servers should be able to extract errors from Kati better, since they look like the errors that Ninja used to write. Ninja is no longer directly connected to the terminal, but its output is read via the protobuf frontend API, so it's just another tool whose output becomes merged together. multiproduct_kati loses its custom status routines, and uses the common one instead. For outputs: The primary output is the ui/terminal.Status type, which along with ui/terminal.Writer now controls everything about the terminal output. Today, this doesn't really change any behaviors, but having all terminal output going through here allows a more complicated (multi-line / full window) status display in the future. The tracer acts as an output of the status package, tracing all the action start / finish events. This replaces reading the .ninja_log file, so it now properly handles multiple output files from a single action. A new rotated log file (out/error.log, or out/dist/logs/error.log) just contains a description of all of the errors that happened during the current build. Another new compressed and rotated log file (out/verbose.log.gz, or out/dist/logs/verbose.log.gz) contains the full verbose (showcommands) log of every execution run by the build. Since this is now written on every build, the showcommands argument is now ignored -- if you want to get the commands run, look at the log file after the build. Test: m Test: <built-in tests> Test: NINJA_ARGS="-t list" m Test: check the build.trace.gz Test: check the new log files Change-Id: If1d8994890d43ef68f65aa10ddd8e6e06dc7013a
2018-05-18 01:37:09 +02:00
case ninja_frontend.Status_Message_ERROR:
n.status.Error(message)
case ninja_frontend.Status_Message_DEBUG:
n.status.Verbose(message)
Add a unified status reporting UI This adds a new status package that merges the running of "actions" (ninja calls them edges) of multiple tools into one view of the current state, and gives that to a number of different outputs. For inputs: Kati's output parser has been rewritten (and moved) to map onto the StartAction/FinishAction API. A byproduct of this is that the build servers should be able to extract errors from Kati better, since they look like the errors that Ninja used to write. Ninja is no longer directly connected to the terminal, but its output is read via the protobuf frontend API, so it's just another tool whose output becomes merged together. multiproduct_kati loses its custom status routines, and uses the common one instead. For outputs: The primary output is the ui/terminal.Status type, which along with ui/terminal.Writer now controls everything about the terminal output. Today, this doesn't really change any behaviors, but having all terminal output going through here allows a more complicated (multi-line / full window) status display in the future. The tracer acts as an output of the status package, tracing all the action start / finish events. This replaces reading the .ninja_log file, so it now properly handles multiple output files from a single action. A new rotated log file (out/error.log, or out/dist/logs/error.log) just contains a description of all of the errors that happened during the current build. Another new compressed and rotated log file (out/verbose.log.gz, or out/dist/logs/verbose.log.gz) contains the full verbose (showcommands) log of every execution run by the build. Since this is now written on every build, the showcommands argument is now ignored -- if you want to get the commands run, look at the log file after the build. Test: m Test: <built-in tests> Test: NINJA_ARGS="-t list" m Test: check the build.trace.gz Test: check the new log files Change-Id: If1d8994890d43ef68f65aa10ddd8e6e06dc7013a
2018-05-18 01:37:09 +02:00
default:
n.status.Print(message)
Add a unified status reporting UI This adds a new status package that merges the running of "actions" (ninja calls them edges) of multiple tools into one view of the current state, and gives that to a number of different outputs. For inputs: Kati's output parser has been rewritten (and moved) to map onto the StartAction/FinishAction API. A byproduct of this is that the build servers should be able to extract errors from Kati better, since they look like the errors that Ninja used to write. Ninja is no longer directly connected to the terminal, but its output is read via the protobuf frontend API, so it's just another tool whose output becomes merged together. multiproduct_kati loses its custom status routines, and uses the common one instead. For outputs: The primary output is the ui/terminal.Status type, which along with ui/terminal.Writer now controls everything about the terminal output. Today, this doesn't really change any behaviors, but having all terminal output going through here allows a more complicated (multi-line / full window) status display in the future. The tracer acts as an output of the status package, tracing all the action start / finish events. This replaces reading the .ninja_log file, so it now properly handles multiple output files from a single action. A new rotated log file (out/error.log, or out/dist/logs/error.log) just contains a description of all of the errors that happened during the current build. Another new compressed and rotated log file (out/verbose.log.gz, or out/dist/logs/verbose.log.gz) contains the full verbose (showcommands) log of every execution run by the build. Since this is now written on every build, the showcommands argument is now ignored -- if you want to get the commands run, look at the log file after the build. Test: m Test: <built-in tests> Test: NINJA_ARGS="-t list" m Test: check the build.trace.gz Test: check the new log files Change-Id: If1d8994890d43ef68f65aa10ddd8e6e06dc7013a
2018-05-18 01:37:09 +02:00
}
}
if msg.BuildFinished != nil {
n.status.Finish()
Add a unified status reporting UI This adds a new status package that merges the running of "actions" (ninja calls them edges) of multiple tools into one view of the current state, and gives that to a number of different outputs. For inputs: Kati's output parser has been rewritten (and moved) to map onto the StartAction/FinishAction API. A byproduct of this is that the build servers should be able to extract errors from Kati better, since they look like the errors that Ninja used to write. Ninja is no longer directly connected to the terminal, but its output is read via the protobuf frontend API, so it's just another tool whose output becomes merged together. multiproduct_kati loses its custom status routines, and uses the common one instead. For outputs: The primary output is the ui/terminal.Status type, which along with ui/terminal.Writer now controls everything about the terminal output. Today, this doesn't really change any behaviors, but having all terminal output going through here allows a more complicated (multi-line / full window) status display in the future. The tracer acts as an output of the status package, tracing all the action start / finish events. This replaces reading the .ninja_log file, so it now properly handles multiple output files from a single action. A new rotated log file (out/error.log, or out/dist/logs/error.log) just contains a description of all of the errors that happened during the current build. Another new compressed and rotated log file (out/verbose.log.gz, or out/dist/logs/verbose.log.gz) contains the full verbose (showcommands) log of every execution run by the build. Since this is now written on every build, the showcommands argument is now ignored -- if you want to get the commands run, look at the log file after the build. Test: m Test: <built-in tests> Test: NINJA_ARGS="-t list" m Test: check the build.trace.gz Test: check the new log files Change-Id: If1d8994890d43ef68f65aa10ddd8e6e06dc7013a
2018-05-18 01:37:09 +02:00
}
}
}
func readVarInt(r *bufio.Reader) (int, error) {
ret := 0
shift := uint(0)
for {
b, err := r.ReadByte()
if err != nil {
return 0, err
}
ret += int(b&0x7f) << (shift * 7)
if b&0x80 == 0 {
break
}
shift += 1
if shift > 4 {
return 0, fmt.Errorf("Expected varint32 length-delimited message")
}
}
return ret, nil
}
// key is pattern in stdout/stderr
// value is error hint
var allErrorHints = map[string]string{
"Read-only file system": `\nWrite to a read-only file system detected. Possible fixes include
1. Generate file directly to out/ which is ReadWrite, #recommend solution
2. BUILD_BROKEN_SRC_DIR_RW_ALLOWLIST := <my/path/1> <my/path/2> #discouraged, subset of source tree will be RW
3. BUILD_BROKEN_SRC_DIR_IS_WRITABLE := true #highly discouraged, entire source tree will be RW
`,
}
var errorHintGenerator = *newErrorHintGenerator(allErrorHints)
type ErrorHintGenerator struct {
allErrorHints map[string]string
allErrorHintPatternsCompiled *regexp.Regexp
}
func newErrorHintGenerator(allErrorHints map[string]string) *ErrorHintGenerator {
var allErrorHintPatterns []string
for errorHintPattern, _ := range allErrorHints {
allErrorHintPatterns = append(allErrorHintPatterns, errorHintPattern)
}
allErrorHintPatternsRegex := strings.Join(allErrorHintPatterns[:], "|")
re := regexp.MustCompile(allErrorHintPatternsRegex)
return &ErrorHintGenerator{
allErrorHints: allErrorHints,
allErrorHintPatternsCompiled: re,
}
}
func (errorHintGenerator *ErrorHintGenerator) GetOutputWithErrorHint(rawOutput string, buildExitCode int) string {
if buildExitCode == 0 {
return rawOutput
}
errorHint := errorHintGenerator.getErrorHint(rawOutput)
if errorHint == nil {
return rawOutput
}
return rawOutput + *errorHint
}
// Returns the error hint corresponding to the FIRST match in raw output
func (errorHintGenerator *ErrorHintGenerator) getErrorHint(rawOutput string) *string {
firstMatch := errorHintGenerator.allErrorHintPatternsCompiled.FindString(rawOutput)
if _, found := errorHintGenerator.allErrorHints[firstMatch]; found {
errorHint := errorHintGenerator.allErrorHints[firstMatch]
return &errorHint
}
return nil
}