A visual interface to work with runtime profiling data from Go programs.

go get github.com/rakyll/gom/cmd/gom

The program you're willing to profile should import the github.com/rakyll/gom/http package. The http package will register several handlers to provide information about your program during runtime.

import _ "github.com/rakyll/gom/http"

// If your application is not already running an http server,
// you need to start one.
log.Println(http.ListenAndServe("localhost:6060", nil))

If your HTTP server is not going to handle the http.DefaultServeMux, you need to manually register the gom handler to respond to "/debug/_gom".

For example, gorilla/mux users can use the snippet below:

import gomhttp "github.com/rakyll/gom/http"

mux := http.NewServeMux()
mux.HandleFunc("/debug/_gom", gomhttp.Handler())
log.Println(http.ListenAndServe("localhost:6060", nil))

Now, you are ready to launch gom.

$ gom

• :c loads the CPU profile.
• :h loads the heap profile (default profile on launch).
• :r refreshes the current profile.
• :s toggles the cumulative sort and resorts the items.
• ↓ and ↑ to paginate.
• :f=<regex> filters the profile with the provided regex.

• Building a lightweight tool that works well with runtime profiles is a necessity. Over the time, I recognized that a lot of people around me delayed to use the existing pprof tools because it's a tedious experience.
• gom has no ambition to provide the features at the granularity of the features of the command line tools. Users should feel free to fallback to go tool pprof if they need more sophisticated features.
• Allow users to filter, hide and ignore by symbol names.
• Increase the awareness around profiling tools and packages in Go.
• Provide additional lightweight stats where possible.

• gom should provide interfaces to let the users to export their profile data and continue to work with the go tool.
• Allow users to work with their custom user profiles.
• Make it easier to generate pprof graphical output.