Use Writer in Context.Status

[willnewrelic]

This PR is an update and fix to #625

If the Context.Writer has been replaced (with something that wraps it), the status code should be sent through the new writer.

[codecov]

Codecov Report

Merging #1606 into master will not change coverage.
The diff coverage is 100%.

@@           Coverage Diff           @@
##           master    #1606   +/-   ##
=======================================
  Coverage   98.92%   98.92%           
=======================================
  Files          40       40           
  Lines        2229     2229           
=======================================
  Hits         2205     2205           
  Misses         12       12           
  Partials       12       12

Impacted Files	Coverage Δ
context.go	99.1% <100%> (ø)	⬆️

---

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 1a1cf65...02d54cc. Read the comment docs.

[chrispine]

Any updates on getting this super-tiny PR merged? Thanks!

[dmarkham]

@thinkerou I do not see any concerns posted here or in #625 about this PR, Do we have a reason it's not part of 1.4 milestone and just get it done?

[yharish991]

is there any update on this pull request?

[willnewrelic]

I have rebased this branch on the latest master. It would love your review!

[dmarkham]

@willnewrelic Realistically until the next release can get approved. Getting things merged/reviewed into master is problematic.

[dangkaka]

can we merge this PR?

[yharish991]

@appleboy can you look at this PR when you get a chance?

[dmarkham]

So after digging in a little deeper the c.Status() looks like it is one of many places that interact with c.writermem directly. I'm afraid making this one change alone will just have things fail in even more strange ways. @willnewrelic you have any samples your testing with you can share?

[willnewrelic]

Hey @dmarkham

I am unsure about the potential for new issues, but I believe the risk is low: Anything that replaces the Context.Writer definitely wants to know what the response code is! The current situation makes instrumentation real tough, and we get issues about it frequently (eg newrelic/go-agent#90).

[paulogustavosouza]

Hi guys, it's possible to merge it ?

[dmarkham]

Hey @dmarkham

I am unsure about the potential for new issues, but I believe the risk is low: Anything that replaces the Context.Writer definitely wants to know what the response code is! The current situation makes instrumentation real tough, and we get issues about it frequently (eg newrelic/go-agent#90).

Oh it's an issue I agree!
Just fixing 1 of 14+ calls to c.writermem spread out in context.go and gin.go seems like we are missing a few of the next issues people will hit. if redirectFixedPath, redirectTrailingSlash, serveError, handleHTTPRequest, ServeHTTP, Copy, and reset Methods are all going to directly use c.writermem I feel like there will be more issues to debug right after this one Status method is fixed by this patch.

What I'm saying is I think this patch is missing a few places and falls short for someone wanting to replace the c.Writer. Fixing just Status is leaving lots of holes if I'm understanding things correctly.
@appleboy @thinkerou have you guys had a chance to think on this?

[namco1992]

Hi just checking if this issue gets any update? thanks!

[chrispine]

We'd love to see this land soon. It's been open for quite a while now. Thanks!

[purple4reina]

What I'm saying is I think this patch is missing a few places and falls short for someone wanting to replace the c.Writer. Fixing just Status is leaving lots of holes if I'm understanding things correctly.

@dmarkham I totally see what you're saying here. If we're going to fix this, we should fix it completely and not just fix one of the many problems that could arise if someone were to replace c.Writer.

What I'm wondering is if we can merge this one part of the fix now, then fix the rest of the issues in a later pull request. This will allow those users who are today replacing c.Writer in order to get instrumentation to start seeing their response codes.

[willnewrelic]

Looks like two people have approved this! Thank you! Perhaps it can be merged ? 😄

[betabandido]

@dmarkham @thinkerou @appleboy Is there any reason to use c.writermem, or can all of its usages be replaced with c.Writer? If there is not, I would love to see this fixed. This issue is also preventing me from using Gin with New Relic to keep track of successful/failed transactions.

[appleboy]

@thinkerou waiting for your review.

[betabandido]

It's great to see this merged :)

@thinkerou @appleboy Are you planning to create a new release anytime soon?