That’s true. I should have mentioned, that there are many test cases and several handlers involved in the real code. I tried to boil it down to a simple snippet reproducing the issue (https://sscce.org/), and some context got lost on that path. Let me explain a little more:
I test a binary (as part of integration tests), which accesses several endpoints on the (OIDC) server. The test cases define proper answers as well as invalid answers to test how the binary handles them. Usually an unexpected answer will let the binary exit, so not all handlers are hit in each test case. Also the server must be the same throughout all test cases, as they build on each other, so I can’t create a new test server in each test case. I thought about using a channel for each handler, but that would not have been stable, as not all test cases hit all handlers and so only the involved handlers would have read from their channel leaving the other channels un-drained. The solution with sync.Map handles this nicely. The key identifies the handler while the value is the response from the test case. For each test case I can write the response to the map, but only the called handlers will actually read from the map and use the value as their response. Unread values is not an issue here compared to the un-drained channels. I guess sync.Map falls under the “Protect the racy data with a mutex or use atomics” category from your list. Thanks Jason Jason Phillips <jasonryanphill...@gmail.com> schrieb am Fr. 31. Jan. 2025 um 20:04: > It's not clear from your example why you're writing to the "response" > variable in the first place. One simple solution would be to create the > test server within the subtest, then there's no sharing at all. > > There are only a few ways to fix a data race and they don't really depend > on the specifics of the application: > * Stop performing concurrent reads and writes (that is, stop racing) > * Protect the racy data with a mutex or use atomics > * Communicate the data over a channel rather than using shared state > across goroutines > > On Friday, January 31, 2025 at 10:22:38 AM UTC-5 Chris Burkert wrote: > >> I meanwhile found a way by using a sync.Map for the responses like this: >> https://go.dev/play/p/KMZ7v63Ht6t >> thanks - Chris >> >> Am Fr., 31. Jan. 2025 um 09:06 Uhr schrieb Chris Burkert < >> burker...@gmail.com>: >> >>> Dear all, >>> >>> I have the following test: https://go.dev/play/p/fQgnvbomlhz (you have >>> to run it locally as it does not work in the playground and it expects >>> /usr/bin/curl as I was not able to reproduce the race using http.Get()). >>> >>> When I run this with *go test -race -v -count=1 .* I get the following >>> race: >>> >>> WARNING: DATA RACE >>> Read at 0x00c0000922e0 by goroutine 10: >>> race_test.TestRace.func1() >>> /Users/D055539/race/race_test.go:16 +0x130 >>> ... >>> Previous write at 0x00c0000922e0 by goroutine 9: >>> race_test.TestRace.func2() >>> /Users/D055539/race/race_test.go:34 +0x44 >>> >>> This is because I write variable response in each test case function and >>> read the same variable in the handler function. >>> >>> My goal is to define the response of the handler for each test case in >>> the table. How can I achieve that without a race? >>> >>> thanks - Chris >>> >> -- > You received this message because you are subscribed to the Google Groups > "golang-nuts" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to golang-nuts+unsubscr...@googlegroups.com. > To view this discussion visit > https://groups.google.com/d/msgid/golang-nuts/a03f0927-2fae-4a9f-99c6-6c505ac45f18n%40googlegroups.com > <https://groups.google.com/d/msgid/golang-nuts/a03f0927-2fae-4a9f-99c6-6c505ac45f18n%40googlegroups.com?utm_medium=email&utm_source=footer> > . > -- You received this message because you are subscribed to the Google Groups "golang-nuts" group. To unsubscribe from this group and stop receiving emails from it, send an email to golang-nuts+unsubscr...@googlegroups.com. To view this discussion visit https://groups.google.com/d/msgid/golang-nuts/CALWqRZrQCx%2Brdcj4oFj2f17B%3DVbZe5NgYdL34sNS7pVY2UkTMA%40mail.gmail.com.
