libURL HTTP downloading

[Phil Davis]

Thanks Trevor! I hadn't thought of downloading in a loop. Good call. I'm 
working with a training system too, where all the media is on a server. 
I really appreciate the feedback (from everyone).

Phil


Trevor DeVore wrote:
> On Aug 8, 2009, at 4:33 PM, stephen barncard wrote:
>
>> Perhaps Trevor could chime in here... he has a upgraded version of 
>> libURL
>> that is used in his GLX Application Framework, however I don't know 
>> if it
>> will work outside his environment or if it can solve your problem, 
>> but it's
>> worth checking out.
>
> The customized version of libURL doesn't change how files are 
> downloaded so there would be no difference in regards to what Phil is 
> trying to do. The customized version of libURL and the GLX App 
> Framework work together to handle interactions with proxy servers but 
> the rest of the features just work:
>
> * Automatic handling of cookies
> * Support for making HEAD requests
> * Callbacks for working with invalid SSL Certs (i.e. display dialog to 
> user to warn them)
> * Non-blocking posts.
>
> On Aug 8, 2009, at 4:25 PM, Phil Davis wrote:
>
>> I assume that, when using Rev's libURL-based HTTP downloading, errors 
>> are more likely to occur when downloading many smaller files than 
>> when downloading a few larger ones. Statistically this makes sense, 
>> but can anyone speak to this from their experience? The "many small 
>> files" scenario I'm talking about is a few hundred files in a single 
>> download sequence, with most files in the under-1mb category except 
>> for a few .mov files that are typically 5 to 70mb each.
>
> I had some training software that would download all of it's media 
> from a server. The amount of media varied from course to course but 
> there was a lot of it. Mostly still images but there were a number of 
> videos as well. libURL worked great. The only thing I remember 
> accounting for was timeouts. I would always download a url in a repeat 
> loop. If there was no error then exit. If the error was timeout then 
> try again, up to 3 - 5 times. I think I picked up this trick from Chipp.
>
> Regards,
>

-- 
Phil Davis

PDS Labs
Professional Software Development
http://pdslabs.net