<p>for t, err = p.token(); err == nil; t, err = p. token() {
...
} </p>
<p>So while t is true and error is nil - perform some stuff then call p.token() again. Correct? Or does it matter if t is true as long as err ==nil </p>
<hr/>**评论：**<br/><br/>nhooyr: <pre><p>As long as err == nil. If err is nil, t is assumed to be ok.</p></pre>ops_man: <pre><p>Thanks for the quick response. </p></pre>RalphCorderoy: <pre><p>You could temporarily break it down into a "while" for understanding.</p>
<pre><code>t, err = p.token()
for err == nil {
...
t, err = p. token()
}
</code></pre>
<p>But personally I dislike the repetition in both, with the risk that they'll diverge, and would use</p>
<pre><code>for {
t, err = p.token()
if err != nil {
break
}
...
}
</code></pre></pre>ops_man: <pre><p>Agreed I like your second example. It's legible and doesn't hurt my eyes to look upon. </p></pre>ops_man: <pre><p>Which is faster? Or is their a difference. Which is to say is it faster to check for nil versus non nil value. </p></pre>RalphCorderoy: <pre><p>Compare the assembly output of the compiler for your target. I'd expect there to be no difference.</p></pre>epiris: <pre><p>Hi, this pattern is represented cleanly in text.Scanner if you are the one designing the API. It places the err check outside the loop. If you are consuming another API it doesn't matter much Ralph has it covered in his post.</p></pre>
