Bojan Smojver wrote:

>On Thu, 2002-08-22 at 09:03, [EMAIL PROTECTED] wrote:
>  
>
>>>>IMO, the best solution is to move the bytes_sent information to the
>>>>conn_rec, and have the protocol module reset it whenever it wants to.  For
>>>>backwards compat, it would be REALLY cool, if the r->bytes_sent could be
>>>>linked to c->bytes_sent, but it can't, so oh well.
>>>>        
>>>>
>>>Why don't we just add up the lengths of headers and the status line and
>>>be done with it? It will be accurate, since we know which headers are
>>>being sent out.
>>>      
>>>
>>It's the body data that I care about, and you don't know that it will be
>>accurate, because the client could have terminated the request half-way
>>through sending the headers.
>>    
>>
>
>OK. I get it now. I hope you guys are not annoyed by my lack of
>knowledge... Sorry, I just started playing with the source a few weeks
>back.
>
>So, bytes_sent will continue to be the amount of body sent out only, no
>headers? If that's the case, then the only option (if we want to know
>total counts) is to introduce bytes_pushed and bytes_pulled (or similar)
>into request_rec. Does that sound feasible to you?
>  
>

That sounds reasonable, although we might end up with some data
from which it's possible to compute bytes_pushed, rather than an
actual bytes_pushed field.  With the c->bytes_sent idea that Ryan
and I were just discussing, we'd probably have to keep track of
something like r->header_bytes_sent in order to properly compute
the content bytes sent for the logger.  So you might end up having
to get add r->header_bytes_sent plus r->bytes_sent to get the total.
But one way or another, I think it's feasible to get the data there. :-)

Brian


Reply via email to