KEN%ORION.BITNET@cunyvm.cuny.edu (Kenneth Ng) (12/22/87)
>From: Glen Overby <ncoverby@ndsuvax.uucp> >Subject: Re: awk/sed (looking for source) > >BitNet has always been a problem. First of all, they run an IBM >protocol which limits lines to 80 characters. To get around this, IBM >has come up with TWO "packetising" techniques ("DISK DUMP and "NETDATA" >to those unfortunate enough to know) which combine short lines and break >up long ones. Unfortunately, mail does not use either of these >techniques, so lines over 80 characters long either get truncated or >wrapped (depending on what software you're using and how it is feeling >today). Actually there are three in major use, DISK DUMP, NETDATA, and CARD. Granted CARD is written outside of IBM, but it is in major use. BTW: you forgot to mention that these protocols allow multiple files to be combined, thus performing the function of SHAR at the same time. Note: the added material in UUENCODE is not that great. It should only be about 25% larger, depending on which version you are using. At least CARD is upward compatible, unlike the umpteen versions of UUENCODE. How do I know? After my last query on sending sources and binaries via mail, I've received about a dozen versions of UUENCODE. And I just have to laugh when I run one version of UUENCODE, and another version of UUDECODE either barfs, or ignores the characters it doesn't understand (Is this what they call a fee_ture (Prounced as a rhyme to 'fee_cee')?) One was even more "clever": it went off the end of the array it kept in memory, and started tiddling with itself. >I feel that the BitNet people should come up with some solution of their >own to make themselves compatable with the rest of the world, rather than >the rest of the world changing to accomodate the IBM world. There is no >reason that all other networks should suffer the increased load of >uuencoded sources or manually translated characters just because of brain >dammage on BitNet. BITNET may be "brain damaged" in a few ways, but at least I can reliably send files and mail back and forth with a very high certainity of them getting through. I've had similar success on ARPANET, but very, very dismal performance on USENET, unless I'm communcating with someone that is directly off a main backbone. >Translation tables have not plagued me with Minix sources, but it still >exists. The problem comes from the lack of a standard EBCDIC to ASCII >translation, as well as internal inconsitiencies in EBCDIC. Curly braces ( >{ } ) are a classic one (There are two sets of them; one for IBM's "TN" >print train and another for the "PN" train. IBM terminals NORMALLY use the >"PN" ones, but many Unix systems use the "TN" one. And a major mail >gateway to BitNet is a Unix machine). Tabs are an even worse situation. >IBM's editors do not use tab characters, but instead display them as a >strange character or something (depending on how you connect to your IBM). >As a result, the gateways expand tabs. Most gateways will expand tabs on a >FOUR character boundary rather than 8. I've rarely had problems with curly braces, its the square braces I have had problems with. IBM has been no help in defining two seperate ones. (Don't think for a nanosecond that I won't flame IBM over something I feel is stupid!) As for tabs, tab characters should never be sent via mail. Read RFC822 first. There is no standard way of expanding tab characters. Some expand to 8, some to 10, others to 4. Some will expand to the next tab stop, some will expand exactly those number of characters. Besides, tab should be a function of the editor, not embedded into the final text. My reference is the typewritter paragon. Does hitting the tab character add a character into the text? It didn't on mine. What it should do is move to the next marked position. >So, in short, the problems with BitNet can be worked around by the people >on that network so rest of us don't have to suffer for it. Or would you >prefer to go back to using punched cards? >Glen Overby >Bitnet: ncoverby@ndsuvax >UUCP: uunet!ndsuvax!ncoverby So, do something about it, write gateway software, write mailers that understand NETDATA formats. And then watch the flames from people that complain about rocking the boat, or doing something different. Especially from those people that think the world is a VAX running BSD4.2 (Note: I do happen to like running programs on a VAX 8550 running Ultrix, which is almost 4.2.) How do I know? Remember a few months back when I said I was writing something that could traverse the major gateways of the world without getting stepped on? Remember all the complaints from people saying that I'm doing something non standard, or that I should use more of the Unix utilities? Your gonna learn that these bigots will complain about anything and everything, but they won't do anything to really change anything for the better.