levine@ics.uci.edu (David Levine) (08/21/89)
Some mailers refuse to post an followup article that has fewer new lines that cited lines. Some netters say that adding filler lines to fake it is "bad", and that a better way is to change to > citation character to something that doesn't get counted as an included line. Why is adding filler lines so "bad"? A few suggestions I've thought of/heard, which don't convince me: 1) waste bits, 2) not as easy to figure out how many filler lines are needed, 3) make printing difficult. -- -- David L. Levine levine@ics.uci.edu or uci-ics!levine Dept. of ICS; University of California, Irvine; Irvine, CA 92717
jgreely@oz.cis.ohio-state.edu (J Greely) (08/22/89)
In article <21376@paris.ics.uci.edu> levine@ics.uci.edu (David Levine) writes: >Some mailers refuse to post an followup article that has fewer >new lines that cited lines. This is a good thing. In general, if you don't have as much to say as the person you're replying to, you haven't accomplished anything. Frequently, you see someone quote an entire article, say "Me, too!", and then insert 30 lines of filler. What have they really said? "I agree, but I have nothing to add to your arguments save my agreement." Surely, the author must then feel blessed. >Why is adding filler lines so "bad"? Because you're adding padding to make up for having nothing to say. Consider: the entire body of the article is counted, and only those starting with '>' count against you. That means that the "whosaid" line, as well as any blank lines, are counted as original. Without saying a word, you've got a good start. Now all you have to do in most cases is manage a sentence for every paragraph you're quoting. Easy, no? >A few suggestions I've thought of/heard, which don't convince me: >1) waste bits, Let's say you quote 30 lines of text (common). You add a one line reply. Now you have to add 27 lines of filler to post it. And do you add one character at the beginning of the line? No, you add a line that says "inews filler. ignore", because that's how you've seen it done before. 27*21==567 bytes. Doesn't sound like much, does it? Of course, there are about 3000 articles posted every day, and if even five percent of them follow your model, it adds over 83 Kbytes to a day's news. Personally, I think that articles should be run through 'uniq' before being counted. It wouldn't stop the problem, but maybe we could make it difficult enough that people would break down and trim annotations (or, better yet, say something!). Of course, counting >'s, was supposed to do that too... -=- J Greely (jgreely@cis.ohio-state.edu; osu-cis!jgreely)
levine@ics.uci.edu (David Levine) (08/22/89)
In article <JGREELY.89Aug21142110@oz.cis.ohio-state.edu> J Greely <jgreely@cis.ohio-state.edu> writes: >In article <21376@paris.ics.uci.edu> levine@ics.uci.edu (David Levine) writes: >>Some mailers refuse to post an followup article that has fewer >>new lines that cited lines. > >This is a good thing. In general, if you don't have as much to say as >the person you're replying to, you haven't accomplished anything. > >>Why is adding filler lines so "bad"? > >Because you're adding padding to make up for having nothing to say. > I should have said this in my original posting to avoid this type of response: I KNOW that it is a good thing. But, there are unusual circumstances under which it is NOT a good thing. My question was why changing the citation character is better than adding filler lines. [filler filler filler filler filler filler] :-)
wisner@mica.Berkeley.EDU (Bill Wisner) (08/22/89)
> Some netters say that adding >filler lines to fake it is "bad", and that a better way is >to change to > citation character to something that doesn't >get counted as an included line. As J said, the best solution is not to quote so bloody much of previous articles. On USENET, brevity is golden. But if you really must quote a huge chunk of someone else's article, and you decide to change the > characters to something else to fool inews, please, at least make sure you don't muck up your References line. Every commonly used UNIX editor I know of allows the user to use regular expressions in search-and-replace commands; so, instead of saying "change > into <" try saying "change ^> into <". (For the uninitiated, a caret at the beginning of a regular expression matches the beginning of a line, so ^> will only match carets at the beginning of lines -- not those at the end of message IDs.) In vi, you can use :g/^>/s//</ (incomprehensible, but effective). In emacs, you can use vi. (cf. alt.religion.computers)
mesard@bbn.com (Wayne Mesard) (08/22/89)
David Levine <levine@wagram.ics.uci.edu> writes: > My question was why changing the citation >character is better than adding filler lines. Because adding filler lines makes the article longer and thus more expensive to ship around the net, store on disk, etc. In short, for the same reasons that these morons have been frothing about .signature files lately. Changing the quote character from ">" to something else doesn't change the size of the article. [Not to mention the fact that it takes no extra effort on the reader's part to deal with a different quote character. The same can't be said for filler lines which one must explicitly wade through.] Disclaimer: Did I say "morons?" What I meant was "concerned participants." -- unsigned *Wayne_Mesard(); "Brain and brain! What is brain?" Mesard@BBN.COM BBN, Cambridge, MA -Kara, Leader of the Eymorgs
charlie@mica.stat.washington.edu (Charlie Geyer) (08/22/89)
In article <21376@paris.ics.uci.edu> levine@ics.uci.edu (David Levine) writes: > Some mailers refuse to post an followup article that has fewer > new lines that cited lines. Some netters say that adding > filler lines to fake it is "bad", and that a better way is > to change to > citation character to something that doesn't > get counted as an included line. > > Why is adding filler lines so "bad"? It's not. The requirement is stupid. Gigalines of garbage are posted every year, but this often silly requirement is the only one enforced by software. It is quite all right to post a 10 line quote with a 1 line response. You should, however, use filler that is blank lines (one character per) rather than cute sayings.
tneff@bfmny0.UUCP (Tom Neff) (08/22/89)
Sometimes the person you're following up to can't seem to say ANYTHING in less than 25 lines (logorrhea) -- if you're pithier, inews can penalize you. When this happens, chop the daylights out of the quoted material, while trying to preserve the meaning. Or [paraphrase]. Don't add extra stuff just to make room, unless it's only one or two blank lines to satify the silly filter. And don't quote e-n-t-i-r-e articles including signature and everything. That's wasteful. Just quote the text you're responding to. ... surrounded with dots ... -- "We walked on the moon -- (( Tom Neff you be polite" )) tneff@bfmny0.UU.NET
dattier@jolnet.ORPK.IL.US (David W. Tamkin) (08/22/89)
Bill Wisner wrote in <30792@ucbvax.BERKELEY.EDU>: | But if you really must quote a huge chunk of someone else's article, and | you decide to change the > characters to something else to fool inews, please, | at least make sure you don't muck up your References line. In rn you can set the -F switch to supply a different citation character. For example, I have '-F"| "' in my RNINIT and the results are apparent above. The message-ID's in the references and attribution lines still retain their angle brackets and I don't have to worry that an editor command to change the citation character will spoil them. I have also trimmed down Wisner's article to the part to which I am responding and resisted quoting his .signature (a common practice that does nothing but waste bandwidth, except when the .signature is the item under discussion). David Tamkin dattier@jolnet.orpk.il.us {attctc|netsys|ddsw1}!jolnet!dattier P. O. Box 813 GEnie: D.W.TAMKIN BIX: dattier CIS: 73720,1570 Rosemont, Illinois 60018-0813 Voice Mail: +1 312 693 0591, +1 708 518 6769
wisner@mica.Berkeley.EDU (Bill Wisner) (08/23/89)
>I have also trimmed down Wisner's article to the part to which I am >responding and resisted quoting his .signature Signature? What signature?
bnews@nixpbe.UUCP (Martin Boening) (08/23/89)
Well, I guess filler lines are bad because they circumvent a news restriction: to never send postings which include more text from other articles than is written by the followup-writer. If a conrtibution to a discussion is 10 lines, you simply should not include 600 lines of reference (like in those 'Subject: Re: blabla (long)' where the whole blabla gets included). Usually a short summary of the article being followed up to should suffice to make your meaning clear and show what you're refering to. Readers who need to know more can always look at the reference line in the header to find articles containing more about the discussion. If you're answering questions, just include the question itself, don't use the text leading up to it. This will usually be enough to help people see what you're refering to. If you absolutely insist on filler lines, use VERY short text. This minimizes the waste byte count :-) BTW: You'll notice that I didn't include the base article. So, if you don't know what the heck I'm talking about, read the article referred to in the References: line. So long Martin -- Martin Boening, c/o Nixdorf Computer AG, DS-CC2, Paderborn, West-Germany Email: | Phone: (+49) 5251 146155 USA: uunet!linus!nixbur!mboening.pad | Fax : (+49) 5251 146108 !USA: mcvax!unido!nixpbe!mboening.pad |
levine@ics.uci.edu (David Levine) (08/25/89)
Summary of responses to my original question:
Why is adding filler lines worse than changing the citation character?
Suggested possibilities:
1) wastes bits,
2) not as easy to figure out how many filler lines are needed,
3) makes printing difficult,
4) more obnoxious to read,
5) some people look at the number of lines when determining whether or
not to read a posting, and filler lines can throw that off.
These trade-off against sticking with the default citation character ('>'):
1) it's easier,
2) you don't have to worry about mucking up other appearances of it, like
in References,
3) some people look for the default citation character when determining
whether or not to read a posting.
There is an easy way to avoid the issue with rn; use the -F option to
change your citation character.
--
David L. Levine levine@ics.uci.edu or uci-ics!levine
Dept. of ICS; University of California, Irvine; Irvine, CA 92717oliver@athena.mit.edu (James D. Oliver III) (08/28/89)
In article <460@nixpbe.UUCP> bnews@nixpbe.UUCP (Martin Boening) writes: >BTW: You'll notice that I didn't include the base article. So, if you >don't know what the heck I'm talking about, read the article referred >to in the References: line. Okay, so how do you find an article by message-ID, given that Rn lists the articles in numerical order within groups, which apparently has no relation. ____________________________ Jim Oliver oliver@athena.mit.edu oliver%mitwccf.BITNET@MITVMA.MIT.EDU
r4@cbnews.ATT.COM (richard.r.grady..jr) (08/28/89)
In article <13871@bloom-beacon.MIT.EDU> oliver@athena.mit.edu (James D. Oliver III) writes: > >Okay, so how do you find an article by message-ID, given that Rn lists >the articles in numerical order within groups, which apparently has >no relation. As an example, suppose I'm searching back for your article, which has an ID of <13871@bloom-beacon.MIT.EDU> I'll be looking for an article with the line Message-ID: <13871@bloom-beacon.MIT.EDU> in the header. So I use the search command ?ID: *<13781@bloom?hr where the "ID:" is the end of "Message-ID:", " *" means any number of spaces (in case there are more than one space here), "<13781@bloom" is the first part of the ID of the message I'm looking for, the question marks mean to search backwards, the trailing "h" means to search all of the header lines, and the trailing "r" means to search already-read articles also. -- Dick Grady r_r_grady@att.com ...!att!mvuxd!r4
tneff@bfmny0.UUCP (Tom Neff) (08/28/89)
In article <13871@bloom-beacon.MIT.EDU> oliver@athena.mit.edu (James D. Oliver III) writes: >Okay, so how do you find an article by message-ID, given that Rn lists >the articles in numerical order within groups, which apparently has >no relation. In "rn" you can do this two ways. The longer more precise way is to use the back-search key "?" as follows: ?d: <460@nixpbe?rh This will hunt backwards for the article with "Message-Id: <460@nixpbe.UUCP>" in the header. The "r" at the end of the command tells "rn" to search messages marked as already read; the "h" says look in the whole header (without the "h" it would only look at the Subject; with a "a" it would search the article text as well). The quicker less precise way is to hit Ctrl-P, which searches backwards for the previous message (read or unread) with the same Subject line as the current. A few quick Ctrl-P's is usually all it takes to find a referenced article. I recommend that all new users take the time to read the documentation for their chosen newsreader - rn or whatever. You can learn a lot. -- "We walked on the moon -- (( Tom Neff you be polite" )) tneff@bfmny0.UU.NET
newsuser@lth.se (LTH network news server) (08/29/89)
>In article <460@nixpbe.UUCP> bnews@nixpbe.UUCP (Martin Boening) writes: >Okay, so how do you find an article by message-ID, given that Rn lists >the articles in numerical order within groups, which apparently has >no relation. I have a little RN macro for this which i thought i could share. It searches (backwards) for the referenced article (the article with the message-id mentioned last on the "References:" line). Place this in a file .rnmac in your login directory and off you go. -- cut here -- # Search (backwards) for referenced article (bound to "l" in article and # pager mode). By Bengt Larsson, <bengtl@maths.lth.se> l %(%m=[ap]??\^Message-ID\:.*%r?hr^J:l) -- cut here -- -- Bengt Larsson Internet: bengtl@maths.lth.se SUNET: TYCHE::BENGT_L
mikeh@dell.dell.com (Mike Hammel) (08/30/89)
In article <14600@bfmny0.UUCP> tneff@bfmny0.UU.NET (Tom Neff) writes:
-I recommend that all new users take the time to read the documentation
-for their chosen newsreader - rn or whatever. You can learn a lot.
I did, but *boy* was it long! I'd like to get a hardcopy so I can study it
periodically without having to telnet over to the news machine (which is
down alot lately) and paging through man rn. Is there a way to print the
man page off? (I asked this on CIS, but I didn't really understand the
answers back then. Maybe I'll understand them better now :-))
Michael J. Hammel | UUCP(preferred): ...!cs.utexas.edu!dell!Kepler!mjhammel
Dell Computer Corp. | Also: ...!dell!mikeh or 73377.3467@compuserve.com
Austin, TX | Phone: 512-338-4400 ext 7169
"I know engineers, they looooove to change things" L. McCoy
Disclaimer: These are my views, not those of my employers. So there.jsdy@dtix.dt.navy.mil (Joseph S. D. Yao) (09/14/89)
In article <3017@dell.dell.com> mikeh@dell.UUCP (Mike Hammel, ) writes: > ... Is there a way to print the >man page off? On a typical machine, manual entries* are stored under directory /usr/man/. If you have an old Unix(R) system or a modern BSD-ische system, manual entries will be stored under /usr/man/manX/topic.Xxxx, where X is the section of the original Unix manual, "topic" is the topic (program, subr, or whatever), and xxx is whatever somebody thought would be cute as a subsection of the Unix manual section. On System III - System V-based systems, these entries are stored under /usr/man/?_man/manX/topic.Xxxx, where ? is one of 'u' (Users' Manual), 'p' (Programmers' Manual), or 'a' (Administrators' Manual). Manual entries are stored in these places in nroff/troff format. This is a text processor format in which formatting commands are inserted in lines before what they are to affect. It is universal to Unix machines (except where vendors try to milk you for more money). The 'man' command typically calls 'nroff -man {filename}' to generate the text that you see. You can do the same, piping it to your print command (lp, lpr, or whatever): nroff -man {filename} | {print command} BUT ... There's a short-cut that 'man' uses on most recent (since 1980 or so) systems. The already-formatted copies are stored in separate direct- ories. On BSD-ische machines, this is /usr/man/catX. On System III++, it is /usr/catman/?_man/manX. Since the text formatting programs take a lot of real and CPU time to run, you might check whether the manual entry already exists in formatted form, and then print that file directly. Word of warning: the formatted form may contain some odd control codes that are supposed to do clever things on specific types of printers. Check with your system administrator whether there is a specific device type name you need to give to 'nroff', or (better yet) whether your printer should be run with 'troff' (or the shells that call 'troff', such as 'xroff', 'psroff', etc.) to get truly typeset-quality output. If not, then you should print a couple of shorter manual entries - or even find whether there are still any manual "pages" to print - and check them for odd happenings before printing the formatter output. If there's anything with which you can't live, you'll have to copy the file into your home directory, and edit it to remove control codes (and make sure all pages are then the length of your printer page - typically 66 lines) before printing it. Have fun. But in moderation. I lost my first Unix account and some of the system staff's trust when I printed out the entire Unix manual at the end of the semester. (But, then, that was only Version 5 or 6 - and you could do that with much less paper.) Joe Yao Hadron, Inc. courtesy David Taylor Research Center * They're much longer than a page, typically; why call them manual pages? (R) Unix is a registered trademark of AT&T Bell Laboratories. PS Sorry about the length ... you wonder why I don't respond so much any more?