<p>I have an API service that at some point should make an http request to an API service that checks access rights. Is there any point of making that call concurrently in contrast to simply making a call?</p>
<p>For example, </p>
<pre><code>func VerifyAccess(someParam string) (bool,error) {
//Should I make call in another go routine ?
// Like this for example
done := make(chan bool)
go func() {
makeHttpCall()
done <-
}
<-done
/*
some other actions here that can't proceed without the response from the action above
*/
// Or should I just make regular call and proceed depending on it state
// Like this
makeHttpCall()
/*
Proceed to do other actions that depend on the result from makeHttpCall()
/*
}
</code></pre>
<p>My main questions is whether there is actual performance benefit in using concurrency when the very next operation depends on the result of the concurrent operation?</p>
<hr/>**评论：**<br/><br/>theclapp: <pre><blockquote>
<p>whether there is actual performance benefit in using concurrency when the very next operation depends on the result of the concurrent operation</p>
</blockquote>
<p>No. Just make the call synchronously.</p></pre>d3sire: <pre><p>Thanks. That was my feeling.</p></pre>theclapp: <pre><p>So the obvious question is, why were you even in doubt?</p></pre>tylermumford: <pre><p>Concurrency is sometimes seen as a guaranteed way to increase performance. :shrug:</p></pre>mre12345: <pre><p>I'm not OP but I have a follow up question. What if two people send a request to a specific endpoint at the exact same time? Won't one request have to wait for the other?</p>
<p>For example if I had this server in node:</p>
<pre><code>const express = require('express');
const sleep = require("sleep");
const app = express();
app.get("/", function() {sleep.sleep(5); res.body = "finished"});
app.listen(8888);
</code></pre>
<p>Then if I send two concurrent requests, the first request will show a response in 5 seconds and the second request will show a response in 10 seconds. </p></pre>djherbis: <pre><p>At least in Go, each request is handled by its own goroutine, so requests don't block one other (unless you explicitly block them with a lock or anther sync. method).</p>
<p>Each of the requests in question would only block 5 seconds from the time of request.</p></pre>mre12345: <pre><p>Thanks for the clarification.</p></pre>theclapp: <pre><p>To expand a tiny bit on djherbis's reply: the OP is talking about sending a request, you're talking about handling a request. Opposite sides of the coin.</p></pre>
