Basic libURL bug?

[Chipp Walters]

Dave,

BTW, I'm using REv 4.0 and libURL 1.1.6. The ulDoProcessLength handler is
not the same as the one you show. Where should I put the delay?

On Sun, Jan 2, 2011 at 9:11 PM, Chipp Walters <chipp at chipp.com> wrote:

> Dave,
> Per your advice, this seems to work more often, but not always. I'll try
> the delay next.
>
> on mouseUp
>    put ""
>    set the httpHeaders to "Connection: close"
>    put URL "http://www.widgetgadget.com/stuff/test.log" into fld 1
>    put the number of lines in fld 1
> end mouseUp
>
> On Sun, Jan 2, 2011 at 8:59 PM, Dave Cragg <dave.cragg at lacscentre.co.uk>wrote:
>
>>
>> On 3 Jan 2011, at 02:27, Chipp Walters wrote:
>>
>> > Dave,
>> >
>> > It's just a standard IIS logfile from a server. There's nothing funny
>> about
>> > it. In fact, I cut/pasted it into a text file and uploaded it to the
>> server
>> > it's on right now, so as to keep Chris from hollering at me! (It had the
>> > problem on our server as well).
>> >
>> > Do you suggest I add the debug line you mention, or is there something
>> > 'better' to do for now?
>>
>> Not sure, Chipp. It's not clear whether we're all seeing exactly the same
>> thing or not.
>>
>> Two suggestions for now:
>>
>> 1.  In the libUrl script where I added the DEBUG 1 line, add this instead:
>>
>>    wait 15 milliseconds with messages
>>
>> This works here. But it fails if I set it to 10 milliseconds. (But I guess
>> this will slow the total download by a number of seconds, so it's a bit of a
>> desperate fix.)
>>
>> 2.  set the httpHeaders to "Connection: close"
>>
>> I was seeing socket errors sometimes (perhaps similar to what Mark
>> reported) when retrying the script. I think the error on the socket probably
>> occurred on the previous try, but for some reason, the error isn't being
>> seen by LC and it is trying to re-use the faulty (or already closed) socket
>> on subsequent calls. Setting the connection header will force the socket to
>> get closed on completion and LC will automatically open a fresh socket for
>> subsequent calls. So not a fix as such, but it may help to reduce some of
>> the consequential effects.
>>
>>
>> Dave
>> _______________________________________________
>> use-livecode mailing list
>> use-livecode at lists.runrev.com
>> Please visit this url to subscribe, unsubscribe and manage your
>> subscription preferences:
>> http://lists.runrev.com/mailman/listinfo/use-livecode
>>
>
>
>
> --
> Chipp Walters
> CEO, Shafer Walters Group, Inc.
>
>


-- 
Chipp Walters
CEO, Shafer Walters Group, Inc.