On 4/13/2011 00:07, michael.vancann...@wisa.be wrote:
<...snip...>
On Mon, 11 Apr 2011, ABorka wrote:
OK, found the problem:
fpweb does not set the content length properly when there was an
exception or error message, it uses the content length from the action
event handler.
For example, if we have
===============
procedure TFPWebModule1.DefActionRequest(Sender: TObject; ARequest:
TRequest; AResponse: TResponse; var Handled: Boolean);
begin
// Handled := true; //commented out to trigger the exception error page
AResponse.Content := '<html><body>Yo! (F)CGI works!</body></html>';
end;
===============
fpweb will send the content length in the response headers as the
length of the string set in the event handler which is the length of
'<html><body>Yo! (F)CGI works!</body></html>'#13#10 in this case, and
not the length of the error response page it constructs and sends back.
Headers>
Status: 200 OK
Content-Length: 45
Content-Type: text/html
This is happening on Windows XP 32bit for me for both CGI and FCGI
applications if the above example is used.
I would not be surprised if this problem would be there in Linux too,
maybe it is just properly displayed in the browser there, no matter
what the header says.
Probably it more depends on the configuration of the Webserver.
As far as I know, the 'content length' header is purely optional.
In each case, I'll patch fpWeb so it sets the content length (if possible)
Michael.
You might be right. After some more testing, the content length does not
seem to be consistent for some reason.
So far, I was not able to figure out why the response is cut short in
the client browser, and when. The received content-length header is
definitely not the length for the response.content in most cases.
I will do some more testing as time permits, first maybe just remove all
content-length headers from the responses or something similar.
I was also trying to see if deflate/gzip-ing the responses on the server
side by Apache makes any difference, but so far no conclusive results
found. The received result pages are just cut short.
AB
_______________________________________________
fpc-devel maillist - fpc-devel@lists.freepascal.org
http://lists.freepascal.org/mailman/listinfo/fpc-devel
