Re: ftpget is adding cr nl to the end of the lines

From: Don Lewis <>
Date: Fri, 10 Oct 1997 05:02:46 -0700

On Oct 10, 1:25pm, wrote:
} Subject: Re: ftpget is adding cr nl to the end of the lines

} My "client" isn't a browser, it's an Alex server (makes FTP archives
} visible as an NFS filestore) that attempts to share cached content with the
} Squid cache.

Sounds pretty interesting.

} The reason it cares about length is that it gets the bytecounts directly
} from the FTP server, and then attempts to get the content through Squid
} (ftpget didn't give very exact bytecounts).
} Then, of course, when the two don't fit, it is SOL.

Just curious -- does it always go back to the FTP server to get the length
even when it will then get a Squid cache hit on the file?

} When I get things from a badly typed system like FTP, I very much prefer
} getting the bytes as stored, and then mangling them using other tools
} locally, to a situation where the network is "helpful" - so what I want
} is a way to tell the network components like SQUID and the FTP server to
} just not try to "help" me. The ;I modifier would seem ideal - and since
} the "client" is a server, not a human, there's no problem with the
} user interface.

Yes, it sounds like modifier support would be ideal for your needs. It
should be pretty easy to add it to Squid. Be aware, though, that there
may be old and obscure systems out there that do ugly things to text
files if you transfer them in binary mode.

It would be nice if Squid was smart enough to know that a ;I URL maps to
the same file in cache as the URL without the modifier if Squid thinks
the file type indicated by the suffix should be fetched in binary mode.
This probably means that Squid should internally add missing modifiers
to ftp URLs.

} It's a complex world :-)


                        --- Truck
Received on Fri Oct 10 1997 - 05:08:39 MDT

This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 16:37:16 MST