At 2002-03-10 10:08 AM +0100, Ulrich Wisser wrote:
>Therefor I would like to suggest to set a "Client-Error"
>header in case libwww sets the error code. Already
>a "Client-Date" is set to mark the time the request was
>finished, so I guess a "Client-Error" would be consitent
>with that and provide additional information.
>
>I would volunteer to make the changes to LWP::UserAgent
>LWP::Protocols::http and LWP::Protocols::https.
>
>Would do you think? Pros /Conns?


I see no cons to your proposal.

But I think it might involve looking over all of LWP for everywhere that an 
error response is produced, to decide whether it should be considered 
client-caused or server-caused.

As to interface: I'd be happy with just the current interface plus a 
"is_client_error" method on HTTP::Response objects.  As opposed to a 
separate Client-Error header, which I don't think I'd have occasion to use.


--
Sean M. Burke  [EMAIL PROTECTED]  http://www.spinn.net/~sburke/

Reply via email to