home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.dcom.modems:13102 comp.mail.uucp:1769
- Path: sparky!uunet!airs!ian
- From: ian@airs.com (Ian Lance Taylor)
- Newsgroups: comp.dcom.modems,comp.mail.uucp
- Subject: Re: UUCP 'G' protocol packet size
- Message-ID: <5259@airs.com>
- Date: 8 Sep 92 04:45:15 GMT
- References: <1079@bazooka.amb.org> <kouqqB4w165w@zswamp.UUCP>
- Sender: news@airs.com
- Followup-To: comp.dcom.modems
- Lines: 51
-
- geoff@zswamp.UUCP (Geoffrey Welsh) writes:
-
- >ti@bazooka.amb.org (Ti Kan) writes:
-
- >> So the question is, is there any benefit in increasing
- >> this to something like 4096 bytes? Do I run the risk of making my
- >> uucico incompatible with other systems?
-
- > First of all, you're confusing UUCP-g with UUCP-G protocols. The larger G
- >protocol was designed to allow varied window and block size, while g protocol
- >is usually implemented with, as you observed, fixed block size.
-
- This is true from a certain point of view, but I think it needs
- clarification. In my experience, most UUCP 'g' protocols support any
- window or block size, when requested by the remote system. The ones
- which don't are the exception. The real difference is that the SVR4
- 'G' protocol provides an easy way to change the window and block size
- which your system requests, whereas most 'g' protocol implementations
- do not.
-
- If your 'g' implementation lets you choose the window and block size,
- you can often increase it with no difficulties, depending on what is
- running on the remote system.
-
- Moreover, every 'g' implementation that I am aware of will accept
- varying block sizes in a single conversation, up to the requested
- limit. The SVR4 'G' implementation will not; every block sent must be
- of the same size, which is surprisingly bad when using large block
- sizes.
-
- > Increasing the block size will increase the window size, and that will help
- >increase throughput where significant latency is involved. However, it is not
- >compatible with most uucicos which don't support G protocol and have an
- >inflexible g protocol block size.
-
- I believe that the latter is a limited set, and that it does not
- include any version of HDB UUCP.
-
- >It's also incompatible with Telebit (and
- >probably Microcom and others) spoofing, which will do more for throughput than
- >larger blocks because they effectively turn UUCP-g into a streaming protocol
- >where it counts.
-
- You are normally better off spoofing if you have a modem which
- supports spoofing, but I believe you can do better than Telebit 2500
- spoofing by making a V.42bis connection with a 1024 byte block size.
- --
- Ian Lance Taylor ian@airs.com uunet!airs!ian
- First person to identify this quote wins a free e-mail message:
- ``He caught the lady in this way: so when the lady was running away for her
- life, he hastily ran to her front and stopped her as a log of wood.''
-