[Gambas-user] [SegFault 11] Enumerating HttpClient.Headers with For Each

[Benoît Minisini]

Le 23/11/2014 22:07, T Lee Davidson a écrit :
> Okay, I've done more testing related to the noise I accidentally sent to
> the list yesterday.
>
> Attempting to enumerate through HttpClient.Headers with For Each causes
> a SegFault 11 under certain conditions.
>
> In the attached project, there is a procedure that traverses through a
> list of three URLs from a GridView and does HttpClient.Get on each one.
>
> If HttpClient.Headers for each URL is enumerated with Headers[iIndex], I
> can run through the list of URLs multiple times all day long. But, if I
> enumerate the Headers array with For Each, the application SegFaults at
> the second execution of the procedure.
>
> When using Headers[iIndex], Headers.Count is correct.
>
> But, when using For Each, URL[1].Headers.Count is obviously
> URL[0].Headers.Count plus URL[1].Headers.Count. And, actually printing
> out the headers shows that URL[1].Headers is the URL[0].Headers with
> URL[1].Headers appended.
>
> Interestingly, the Headers for the third URL are correct no matter which
> enumeration method is used. And, the application does not SegFault when
> attempting to enumerate the Headers (with For Each) for the third URL on
> the first run. Only after the procedure is re-entered the second time
> does the app SegFault.
>
>
> How to reproduce:
>
> Set "Actually Print Headers to Console" to your preference.
>
> A.
> 1. Leave "Enumerate Response Headers With" set to "Headers[iIndex]".
> 2. Click "Get Response Status Codes" as many times as you wish to show
> that enumerating with Headers[iIndex] does not SegFault.
>
> B.
> 1. Set "Enumerate Response Headers With" to "For Each Headers".
> 2. Click "Get Response Status Codes" twice. (>>SegFault)
>
>
> One last note: I stumbled across this while trying to determine why
> HttpClient.Code sometimes returned "0", and IIRC, I was enumerating with
> For Each at the time.
>
>
> Now just for clarification, is enumerating HttpClient.Headers with For
> Each something I should not be doing? Have I done something else wrong?
> Or is this actually a bug?
>

I will look at it.

But if you have a segfault, this is a bug, whatever the reason. You are 
not doing anything wrong, the interpret should never crash, only raising 
errors.

If the crash is a library bug, I can't do nothing, but it seems to not 
be the case there.

Regards,

-- 
Benoît Minisini