On Fri, Aug 11, 2017 at 4:38 PM, Ian Lance Taylor <i...@golang.org> wrote:
> On Thu, Aug 10, 2017 at 1:43 PM, <spen...@justin.tv> wrote:
>>
>> Yes, it makes sense that it would be impossible to really check at that
>> level. What surprised me was that this does not trigger vet:
>>
>> for i := range slice {
>> go f(i)
>> _ = 1
>> }
>>
>> Yet this does trigger vet:
>>
>> for i := range slice {
>> _ = 1
>> go f(i)
>> }
>>
>> Is there something special about the last statement that makes it easier to
>> check?
>>
>> For what it's worth, this exact bug caused real user-impacting issues at my
>> company. The team responsible for the code was very surprised that vet did
>> not alert them to the issue before it made it into production.
>
> I agree that this appears to be a bug. I filed
> https://golang.org/issue/21412.
If you read the issue, you will see that it is clearly more subtle
than I initially thought. If you have examples that vet could
reliably catch that it does not catch today, that would be helpful.
What did the code look like that had a uncaught case that surprised
people? Thanks.
Ian
--
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.
For more options, visit https://groups.google.com/d/optout.
