--- In the_gdf@y..., "jbt00000" <junkmail@j...> wrote:
> Are you only including members in the mesh whose Hashes match, or
> any that Bearshare deems equivalent?
Any that BearShare deems equivalent. Note that 2.5.0 groups files
into two categories: those with hashes, and those without. Therefore
hash files will always list sources that promise a hash.
You can tell if a requestor obtained the mesh information via hashes,
since the requestor will provide the X-Gnutella-Content-URN header.
You can tell if the provider's alternate locations refer to hashed
content by the presence of the X-Gnutella-Content-URN header in the
HTTP response.
> If it is the latter, how are you handling servents that don't
> support url encoding (i.e. Limewire). I would suggest not inluding
> them in the mesh, as servent using the mesh data wouldn't have
> enough hints to make those uri's work.
I don't see a problem. LimeWire doesn't support URL encoding in the
HTTP request. But then again, no one supports the reading and parsing
of the X-Gnutella-Alternate-Location header.
When someone implements the handling of the header, they will have to
url decode the URI and submit a non url-encoded HTTP request to the
appropriate servent.
The separate problem of not handling url encoded URIs on HTTP
requests is solved by first releasing widespread support for url
encoding (BearShare already supports it, and LimeWire will soon),
then making the change at the request level.
> What versions of Bearshare supports the proper Base-32 alphabet?
BearShare 2.4.2 and later support the Base-32 alphabet described in
HUGE 0.93
In case you didn't know, this is the format of the packed private
data block in BearShare query hits:
unsigned char flags;
unsigned long version;
unsigned long uptime; // in minutes
//...
};
The version would be 0x00020402 for BearShare 2.4.2.
Our solution to the problem of having two Base-32 formats is to first
try to identify the Base-32 version by looking at the characters
used. This can produce only positives and false negatives. If it is a
negative then we resort to checking the version number. If there is
no version number (i.e. not a BearShare servent) we assume it is HUGE
0.93 (since no one else is sending hashes yet).
Are you only including members in the mesh whose Hashes match, or any that Bearshare deems equivalent? If it is the latter, how are you handling servents that...
... Any that BearShare deems equivalent. Note that 2.5.0 groups files into two categories: those with hashes, and those without. Therefore hash files will...
... or ... Therefore ... hashes, ... header. ... the ... "X-Gnutella-Alternate-Location" should ONLY be used when there is also a "X-Gnutella-Content-URN"...
... inluding ... the ... parsing ... to ... the ... This will work assuming that all servents will continue to support both url encoded and non-url encoded...
... Damn it Vinnie, cut the crap! What YOU want is not what EVERYONE ELSE wants! And some like to discuss possible alternatives, so we can find the solution ...
... That's OK. I'll stick to the official HUGE documentation, using the RFC822/1123 format as noted in section 6.2.2, since I can reuse the code for the...
Mike Green wrote:
> I need more coffee....
Maybe a cup of Java? ;-))
SCNR _________________________________________________________ Do You Yahoo!? Get your...
... .. ... Come to think of it, those clients using the X-Live-Since handshake header could reuse the date/time parsing code for this purpose. But that'd be ...
Hi ... How about making it easy for the client by giving it the information it wants: The age of the Alternate-Location. This way the client does not need to...
(repost) ... That's OK. I'll stick to the official HUGE documentation, using the RFC822/1123 format as noted in section 6.2.2, since I can reuse the code for ...
... Gojomo, can you please update HUGE to use YYYYMMDDHHMMSS for the date/time format? I think there are still some GDF members under the misconception that...
... the ... the ... I for one would have preferred to use an existing standard, but am happy to go along with Vinnie's word from God. Why? Since the form in...
Your YYYYMMDDHHMMSS format looks fine to us. We will use that unless there is a total rebellion (which there shouldn't be). Thanks -greg ... From: freepeers...
... Unfortunately, you're not allowed to do that. This does not conform to the HTTP/1.1 standard date representation. Recall that the downloading mesh could...
... You might question the way it was pushed but at least we're getting some resolution to the issue. ... Yeah but there was a little rumble about parsing the...
... Yes, it is I who found that parsing problem. It is due to the fact that: X-Foo: a X-Foo: b Can be rewritten as: X-Foo: a, b However, if one replaces the...
... unless there ... Gojomo recommends adding a few extra characters (YYYY-MM-DD-HH:MM:SS I believe) for one reason or another. Since we haven't actually...
The 802.11 protocols apparently include peer to peer capabilities that are beginning to be deployed. This will tend to make 802.11 clients (laptops, PDA's)...
The book, "Ad Hoc Networking" by Perkins is a good reference on wireless routing protocols. ... -- Justin Chapweske, Onion Networks http://onionnetworks.com/...
Justin Chapweske
justin@...
Mar 7, 2002 6:32 pm
I think that wireless P2P networks are really going to explode in the near future as the technology advances. I envision being able to access the internet for...
steve bryan wrote:
> The 802.11 protocols apparently include peer to peer capabilities
> that are beginning to be deployed. This will tend to make 802.11
>...
I hope we are going to have one format and not three allowable date formats. Its all well and good that HTTP/1.1 supports three but that doesn't mean that we...
Standards-reuse is good, as are ISO8601 dates. But when HUGE was first authored, I couldn't find an authoritative reference on a canonical ISO8601 date+time ...
... Just because a few people had an elucubration in 1998 does not mean we must have the same in 2002. You'll have noted that the mentionned TR from W3 was...
Raphael_Manfredi@...
Mar 7, 2002 11:18 am
Raphael_Manfredi@... wrote:
> And HTTP/1.1 says that:
> > HTTP-date = rfc1123-date | rfc850-date | asctime-date
> > This should put an end to this...
... Sorry, but discussing about the date format in an HTTP header IS insane. Therefore, I'm not even discussing it, I'm pointing to facts. The point being...
... This is a VERY BAD idea. Don't do it! First, it's not readable at all. You must keep HTTP headers readable. The mere fact that no human usually reads...
... Because Jason does not remember that I pointed out that it was enough to replace the ", " between the URL and the date with a "; ", since a semi-colon...