Re: pipelining setting

On 6/29/13, Robert Max Jackson <[email protected]> wrote:
>    Thanks for the quick responses.
>
>     Yes, I normally assume "1" is on and "0" is off.  What I could not
> figure out is how the documentation for "tolerate-pipelining" conformed to
> that.
>
>     As the initial response (Lee) notes, the documentation says the default
> value is "None".  But the description implies the default value is to not
> tolerate pipelining, which would presumably mean the default value is "0".
> However, (at least with the Windows installation package, which I use), the
> final lines in this section as installed are:
> #  Examples:
> #
> #      tolerate-pipelining 1
> #
> tolerate-pipelining 1
> #
> That is, it explicitly sets "tolerate-pipelining" to "1".  Since that is in
> the "config.txt" as installed, that is the default, right?

The default is what Privoxy uses if the keyword is not specified.  In
this case, 0.

>     Hopefully, this clarifies why I find the documentation ambiguous.
> Perhaps it was just an error to include the line that sets it to "1".

I thought it was an "improve the user out-of-box experience" thing.
My recollection is that tolerate-pipelining was added because Opera
would use pipelining even if the configured proxy didn't support it.
So having config.txt enable it means Opera users don't have to change
anything to have Opera work.  But yes - I can see how the doc saying
the default is x and config.txt setting it to y could be confusing.

>     As to improving the documentation, the first need is to make it
> consistent I believe.  Beyond that, it is always clearer if the
> documentation simply lists the possible values and their effects, as was
> done in the response email, e.g. (assuming the directions are correct)
>    0: Privoxy rejects pipelined requests
>    1: Privoxy tolerates pipelined requests
> I would recommend avoiding the word "unset" which is inherently ambiguous (a
> paramenter is always "set" if given any value),

good  point

>    substituting something like:
> "Effect if set to '0' (pipelining-tolerance turned off): ...".
>
>     Alternatively, both possibilities for altering the documentation in
> Fabian's note (received after writing the above) seem sensible and effective
> to me.
>
> Thanks again .... Robert Jackson
>
> P.S.  My apologies beforehand if I am replying incorrectly.  I am not
> familiar with the way this kind of listserve works.

You're doing fine :)

Regards,
Lee


>
>>>> Fabian Keil <[email protected]> 6-29-13 1:19 PM >>>
> Lee < [email protected] > wrote:
>
>> On 6/28/13, Robert Max Jackson < [email protected] > wrote:
>> > For Privoxy version 3.0.21, after rereading sections 6.5 of
>> > config.txt on tolerate-pipelining repeatedly, I cannot figure out
>> > which setting, 0 or 1, tolerates pipelining. (I fear the
>> > documentation here is completely ambiguous, at least for those not
>> > familiar with whatever linguistic assumptions would make it
>> > unambiguous.)
>> >
>> > Could someone clarify?
>>
>> 0: privoxy does not handle pipelined requests
>> 1: privoxy attempts to handle pipelined requests from the client (eg.
>> firefox)
>>
>> # Effect if unset:
>> #
>> # If Privoxy receives more than one request at once, it
>> # terminates the client connection after serving the first one.
>>
>> I'm so used to 1 is the same thing as "set" and 0 is "clear" or
>> "unset" that it's hard for me to imagine someone not knowing that --
>> assuming that is the problem with the documentation.
>>
>> How would you change the documentation to make it clear that
>> 0 = client pipelining is not allowed
>> 1 = client pipelining allowed
>
> I suppose we could change "This option lets Privoxy tolerate pipelining."
> to something like "Setting this directive to 1 lets ...".
>
>> Fabian - is this bit correct?
>>
>> # Default value:
>> #
>> # None
>>
>> sure looks like the default is 0
>
> In the config file "None", "Unset" and "0" currently refer to the same
> numeric default value. I wouldn't be opposed to consistently using "0".
>
> Fabian
>
>

------------------------------------------------------------------------------
This SF.net email is sponsored by Windows:

Build for Windows Store.

http://p.sf.net/sfu/windows-dev2dev