[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[leafnode-list] Re: Requesting "legal opinion" on Newsgroups & Xref headers
On Sat-2009/11/21-20:38 Whiskers wrote:
> Found it :)) It's a FAQ in six parts, and the latest posting was
> Date: 12 Nov 2009 05:19:14 GMT
> Expires: 26 Dec 2009 05:18:16 GMT
> Message-ID: <internet/info-research-faq/part1_1258003096@xxxxxxxxxxxx>
> Message-ID: <internet/info-research-faq/part2_1258003096@xxxxxxxxxxxx>
> Message-ID: <internet/info-research-faq/part4_1258003096@xxxxxxxxxxxx>
> Message-ID: <internet/info-research-faq/part5_1258003096@xxxxxxxxxxxx>
> Message-ID: <internet/info-research-faq/part3_1258003096@xxxxxxxxxxxx>
> Message-ID: <internet/info-research-faq/part6_1258003096@xxxxxxxxxxxx>
> [on lone ^M characters sticking out]
> I can't, not on its own; all six articles have the usual hex 0D 0A
> characters at the end of the Newsgroups header. So it doesn't look as
> though the problem exists in the original articles, unless it's getting
> corrected somehow on the way to me.
> ^M on its own is the usual Unix and Linux newline - but for some reason,
> DOS and Windows require ^M^J together, and usenet follows that convention.
> So perhaps the file you are looking at has been edited in a program set to
> use the Unix-style newline?
I rather think unix normally uses ctrl-J, hex 0x0A (linefeed), whereas
DOS and Windows use ctrl-M + ctrl-J, hex 0x0D + 0x0A (carriage return
+ linefeed) to end a physical line.
> Only the combination hex 0D 0A (^M^J or carriage return and line feed)
> is allowed, as far as I know. (Although some usenet clients can cope
> with format=flowed as well).
Most internet protocols that operate with a textual representation use
hex 0x0D + 0x0A to end a line _on the wire_. Clients and Servers should
normalize the protocol parts of any data to this convention while
leaving the payload alone. The "Newsgroups" header is upper layer
protocol data used for routing articles, "Xref" is more like house-
keeping data for servers, we throw it away (and make our own) as it's
irrelevant once handed off to a USENET peer.
Unfortunately, Mr. Myers did not say which version of leafnode he is
using and what server sent that article.
Matthias tries to make sure his products respect the RFCs closely.
I myself cannot verify the original problem. The archive at  doesn't
show anything unusual, but it's an archive, not 1:1 copies of original
Lone ^M's often come from texts handled by Microsoft products, they
stand out on unix systems and can be repaired by sed-scripts or even
auto-converted by tools like zip.
BTW, thanks, Whiskers, for keeping an eye on the list! You often step
in with helpful answers 8-)
leafnode-list mailing list