IgnoreMatch() Option

[ryanrhee]

Option to ignore stack traces that have the full substring inside it. (Can't be done in user-space because the opts struct is private.)

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[codecov[bot]]

Codecov Report

Merging #30 into master will increase coverage by 0.13%. The diff coverage is 100%.

@@            Coverage Diff             @@
##           master      #30      +/-   ##
==========================================
+ Coverage   95.03%   95.17%   +0.13%     
==========================================
  Files           4        4              
  Lines         141      145       +4     
==========================================
+ Hits          134      138       +4     
  Misses          4        4              
  Partials        3        3

Impacted Files	Coverage Δ
options.go	100% <100%> (ø)	:arrow_up:

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update 71b3642...d19a1b8. Read the comment docs.

[prashantv]

Thanks for the contribution @ryanrhee

Can you provide some additional context on what exactly you're trying to do? Ignore a substring match in the full stack is a pretty broad feature, typically exclusions are a little more fine-grained (based on the top function on the stack, or looking for a specific function).

This sub-string match will also match filenames, which can vary across environments, which could have unexpected results.

[ryanrhee]

@prashantv ideally each stack frame would be parsed and i could match against it, but it seems such parsing isn't easily doable in the current codebase. Sub-string match will give us a reasonable way of continuing to run goleak without investing in a way to properly parse stack frames; the alternative for us would be to just stop using it, and we prefer not to do so.

[prashantv]

@prashantv ideally each stack frame would be parsed and i could match against it, but it seems such parsing isn't easily doable in the current codebase. Sub-string match will give us a reasonable way of continuing to run goleak without investing in a way to properly parse stack frames; the alternative for us would be to just stop using it, and we prefer not to do so.

Are you trying to ignore a specific function in the call stack? Will it be the full function name, or is there a use-case for a sub-string? We can ignore the top function already, we could extend the parsing to ignore a function anywhere in the frame. It would be a little more work, but I'd prefer to make this library API safer for the long-term.

We are planning to cut a 1.0 soon, after which we'll maintain the API going forward, so I want to be careful about the APIs we are adding.

[ryanrhee]

Full function name.

On Wed, May 8, 2019 at 2:04 PM Prashant Varanasi [email protected] wrote:

@prashantv https://github.com/prashantv ideally each stack frame would be parsed and i could match against it, but it seems such parsing isn't easily doable in the current codebase. Sub-string match will give us a reasonable way of continuing to run goleak without investing in a way to properly parse stack frames; the alternative for us would be to just stop using it, and we prefer not to do so.

Are you trying to ignore a specific function in the call stack? Will it be the full function name, or is there a use-case for a sub-string? We can ignore the top function already, we could extend the parsing to ignore a function anywhere in the frame. It would be a little more work, but I'd prefer to make this library API safer for the long-term.

We are planning to cut a 1.0 soon, after which we'll maintain the API going forward, so I want to be careful about the APIs we are adding.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/uber-go/goleak/pull/30#issuecomment-490589823, or mute the thread https://github.com/notifications/unsubscribe-auth/AAB3G5CJGTCJAJQAY4MOOVTPUMIZ5ANCNFSM4HJ4DCKA .

[frioux]

I would like to be able to us this at least to ignore the eternal goroutine that glog starts up at init time. (search for init in https://github.com/golang/glog/blob/master/glog.go)

[prashantv]

I think extending the stack parsing code to parse the function names, and then allowing an ignore if a function shows up anywhere in the stack trace is a better API to support, rather than allowing matching against the filename which is arbitrary.

I was hoping to spend some time improving the stack parsing functionality, but did not get around to it, but would appreciate any contributions on that front, thanks!

[abhinav]

Update: There's stack trace parsing code now, and #113 added a new IgnoreAnyFunction option to ignore a function anywhere in the stack trace.

Closing this PR since ignoring a specific function was the goal here too. Thanks for opening the PR and the discussion, @ryanrhee, and thanks for suggestions on how to fix this @prashantv.