[noCensorship] Re: 609 proxies on uncommon ports - checked 2004-01-20
- From: wayne <wayne@xxxxxxxxxxxxx>
- To: freerk@xxxxxxx, nocensorship@xxxxxxxxxxxxx
- Date: 21 Jan 2004 15:49:55 -0000
> From: Freerk <freerk@xxxxxxx>
> To: nocensorship@xxxxxxxxxxxxx
> Subject: [noCensorship] 609 proxies on uncommon ports - checked 2004-01-20
>
> Hello nocensorship,
>
> I'm sorry, but I am very busy this month. The next list will
> come in about mid-februrary, I think.
>
> I always only post the working proxies I have, no matter how
> slow they are or if they are censoring etc.
> Should I test them better?
You probably can't do the 'working'/'not working' test any
better. They die quickly, but you can't predict that. When
I test your lists for the ProxyTools database (just a day or
three after you post), there's usually over 30% dead.
> With which program? BTW, does anyone ne=
> ed
> Wingates?
> To sort them by speed is useless, because on the other side
> of the world the speed is really different...
The slowest ones will be slowest no matter where you use
them from. The only difference is the network latency which
could easily add 1/2 a second to each group of TCP packets
ACKed.
Users with screwed up early version Microsoft TCP/IP stacks,
have a bigger problem of course ...
The effect of network latency is far more obvious when
requests with small responses are made, because of
TCP/IP's slow start algorithm on each new connection.
So...you'll see lots of difference across the world (due to
network latency, user's TCP/IP stacks, bad browsers (that can't
use keep-alive, pipelining, or even compression), but for any
particular user, all those things will be constant and his
relative experience of the speeds of various proxies (roughly
the same 'latency' away) will depend on the actual proxy speed.
Those programs you use probably don't test speed well though.
To really test the user's experience of the speed of a proxy
you need to test a full reference page download, not just a
quick HEAD, or even a short page. I think these programs are
designed to scan large lists of proxies and they make lots of
such compromises to do that in a reasonable time.
Have you never tried statProxy? You need Perl and the latest
versions of the whole proxyTools package from sourceforge (see
my sig below). The released packages are nearly a year old now,
so don't test those. You would only need to use test 0 to get
your lists of working proxies. At the same time, you would get
the real speed and extra proxy information (type, version, etc.).
SP was originally a very slow, but sure, proxy analyzer. Later
versions overlap tests and are much faster (now I call it a
scanner). It doesn't need to compromise tests to gain scanning
speed.
I'd be interested to hear any comments you have.
BTW, test 13 is an anon test (but not a 'judge'), test 18 is a
socks4 test, 14 and 20 are noncensoring tests, etc.
I can add a socks5 test if there's some interest.
perl statProxy.pl -h
lists all the examples and the test descriptions.
> Tested with: AATools 5.50 and Proxy Checker 7.0
> Env used: this Proxyjudges
>
> Best regards,
> Freerk mailto:freerk@xxxxxxx
--
Turing email: wayne at nym.alias.nest
(remove the obvious extra 's')
http://proxytools.sourceforge.net/
===8>============== noCensorship community ===============
List's webpage: //www.freelists.org/webpage/nocensorship
List's archive: //www.freelists.org/archives/nocensorship
To unsubscribe: nocensorship-request@xxxxxxxxxxxxx with 'unsubscribe' in the
SUBJECT field.
Moderator's email: nocensorship-moderators@xxxxxxxxxxxxx
===8>============== noCensorship community ===============
Other related posts:
