April 16, 1990
Note: A rapidly changing environment made this piece obsolete in short order. It is offered here merely as a historical curiousity.
The INTERNET is a loosely affiliated network of computer systems which include military and commercial research centers and academic computer centers; such as are home to big mainframe computers, including super-computers. As such, the INTERNET functions something like the interstate highway system. As a network, it permits users to log on to remote systems across the country. Telecommunications and computer technology have established a national and international system that empowers all users on the network by providing a thoroughfare to remote sites.
For librarians, especially academic librarians with access to the campus computer system, the nearly forty academic library catalogs available become a major resource for reference, cataloging, and collection development. In addition, many libraries provide additional local databases which are useful to others.
The first question you might ask is, given the rising costs of long-distance telephone service, how much are these cross-country connections going to cost. As it turns out, the cost of the long-distance connection is zero; the cost is picked up in the overhead (and grants) of the respective computer centers. And, since the long-distance connection itself uses relatively few local cpu's (computer processing units, which are billed at thousands of dollars per hour) the total cost is probably less than a buck. And the network itself functions to increase cooperation by not billing the remote user for access. Accordingly, it is frequently cheaper to check the catalog at Berkeley than the dial-up port of the local university catalog. In this respect, libraries have not yet attained the higher consciousness evidenced by the computer centers in that many (billed hard dollars by the computing centers) pass on those charges to users. In addition, some special services, such as CARL's Uncover, are only available by subscription and the payment of licensing fees.
Access to the INTERNET is through the institutional mainframe (with an INTERNET portal) which usually requires a USER-ID, PASSWORD, and ACCOUNT NUMBER. Many universities subsidize student and faculty use of the computer by billing against allocations (although considered computer funny-money it is essential for allocating cpu resources) which makes the cost to the individual user, again, zero.
Once equipped with an ID and PASSWORD and having learned the basics of signing on and getting accustomed to extremely awkward user interface (at least for those familiar with PC software), the first step is to open up the link to the network. One version of this command is TCPLNK which probably means something like Transfer Command Protocol LiNK.
Next, issue the TELNET command to address the specific remote host. The form is TELNET <ADDRESS> where <ADDRESS> is one of two forms: the domain name or the ip address. The domain name is in the form MELVYL.UCOP.EDU while the ip address is more like a zip code: 126.96.36.199 (both examples are for the University of California's on-line catalog). Some systems require the <ADDRESS> be in the domain form while others need the ip address. The WHOIS command (as in WHOIS MELVYL.UCOP.EDU or WHOIS 188.8.131.52) should provide the alternate form, although this did not work for me.
This logs you on to a foreign computer which can be accessed just as if it was in your own neighborhood; at least if you've managed to match your hardware and software with their expectations. Clifford Lynch in a recent article on designing library catalogs for INTERNET access provided a glimpse of some of the problems that can arise.
First among these is hardware. The variety of hardware devices used to access computer systems is quite large in part because each technological advance must be evaluated for its effect on the existing user base. In the 1950's the television industry was considering whether to adopt color technology that was consistent with black and white (and thus could be viewed on a black and white set) or to change to a technically superior system that would disenfranchise all b&w sets or require broadcasters to set up two transmitters. The choice was made to maintain backward compatibility over better resolution.
The console interface to a mainframe computer has the same problem. Some users are still using twenty-year old teleprinters to make their connections while others have upgraded to the newest, multi-key, super-resolution display technology. Some are using hard-wired terminals perfectly matched to the host computer and others are dialing in with a PC and generic communications software. As a result, it is not a given that your connection with the remote computer will appear on your screen with any kind of legibility or that you will be able to respond to its prompts.
Some of this is the result of variations in software, particularly if you are using one or another form of terminal emulation. A hardwired terminal offers the most security in matching terminals to systems. Even this is no guarantee if the remote system does not support the terminal you are using. Those dialing in from a PC with communications software are caught in an even stranger circumstance. Not only is the terminal display software dependent, the PC keyboard is usually remapped to a specific emulation. If prompted to hit the PF-7 key, you must know that in your emulation this means to hit ESC-7. Discovering these codes and changes can be something like asking for a restroom in Bulgaria equipped only with a French-English dictionary.
There are two standard software emulations provided in communications programs: ASCII and VT100. ASCII is primarily used for PC-based BBS calling at 8-N-1 (8 data bits, no parity, 1 stop bit) while VT100 is more common calling mainframes which usually are set at 7-E-1 (7 data bits, even parity, 1 stop bit). VT100 is based on a Digital Equipment Corporation standard hardwired terminal.
In the common VT100 emulation in PC communication software Procomm 2.4.2, the terminal function keys (PF1-PF12) have been remapped to <ESC 1>-<ESC => or the top twelve keys. F1 and F2, however, work just like <ESC 1> and <ESC 2>. The VT100 key command CLEAR is issued as <SHFT F8> but can also be replicated in some modes by <ESC .>. This is particularly useful when you see the HOLDING message at the bottom right of the screen. ENTER, a common mainframe command, is not the keyboard Enter or Carriage Return key (at least not always) but <SHFT F10> or <ESC O M> (the letter O and not zero). PF4 is sometimes used as part of a logoff sequence (such as at Texas A&M) and must be entered as <ESC O S>. Little of this makes any intuitive sense and much of it is not clearly documented. It is also useful to log each session to disk (<ALT-F1> in Procomm) for later review.
Another problem with the hardware/software combination is the screen display. Aside from garbage characters ("happy faces from hell," according to one wag) I have encountered invisible menus (BOOTES.UNM.EDU) as well as the more common screen length anomaly. Procomm, as do many communication software programs, reserves the bottom screen line for a status display. Many systems send 23 of 24 lines, prompt for response, clear screen and print one more line, then wait for another response. After twelve screens, there is no relationship between what the designer imagined you would see and what is visible on the screen. At its best this is merely an annoyance; at its worst it becomes impossible to coordinate responses with prompts.
Usually a call to the local university mainframe starts with a request for terminal type to which the response is almost automatically VT100. Using the INTERNET gateway to call another mainframe results in another request that you specify terminal type. Entering VT100 at this point may not be the best choice because of the nature of the transmission of signals through the TELNET structure. I don't know why this is so, but it is. Some remote systems (and probably your local system as well) will list allowable terminal types if you enter a "?" or "HELP" at the prompt. The choices can range from VT100 to TYPETERM to UNKNOWN to TINCAN. Unfortunately, there is no easy way to change it after the session has started; although you might try the command SET TERMTYPE UNKNOWN. Experimenting to find the best one is frequently a matter of signing on, checking how it works, signing off, signing on under another form, and so on. Part of the difficulty is that different systems have different ground rules for terminal types. For example, one system offers a VT100 option as locally modified. Accordingly, its responses can be unexpected. Others seem to offer no alternative save a perfect match. The University of Oregon's JANUS catalog requires users to be either VT100, WYSE or TVI925 terminals. None of these correspond to Procomm's VT100 emulation.
Connections made through a PC modem using VT100 emulation will often find a surprising amount of strange symbols (probably escape codes for the visual display) cluttering up the screen if VT100 is chosen on the remote. In these cases, it is best to choose the lowest common denominator terminal mode if possible. Promising choices include UNKNOWN, TTY, HARDCOPY, TYPETERM, and sometimes IBMPC.
Some of this depends on whether the connection through the INTERNET is made in Line Mode or Transparent Mode. Line Mode appears to be more common but a few systems connect at Transparent Mode. The latter apparently use TN327x terminals which have (as near as I can tell) lots of specially marked keys not available on your PC. The different modes change the nature of the display. They also change TELNET commands which are especially important in logging off and exiting the remote system. Some systems, Michigan State University, for example, offer two addresses; one for line mode and one for transparent mode.
Once logged on the remote system, it should be a matter of just following the prompts, if any. However, the connection might be directly to the library catalog or to some sub-computer at the university. If it is the first, there are usually helpful screens, menus, and clear prompts as most are designed to accommodate undergraduate users. However, if it is the latter, it might be necessary to wade through a jungle of confusion to reach the relative oasis of the library catalog. For instance, the University of Kansas sequence is as follows:
At "Username:" enter "RELAY" <CR>
At "System?" enter "OCAT" <CR>
At "Terminal Type?" enter "?" for a list of acceptable types.
At "Enter transaction" enter "OCAT" <CR>
Just as it is necessary to step through (or log on to) successive layers of the system, it is also important to back out through each layer in logging off.
This is not always easy. Logging off of any remote system sometimes seems impossible. Although a few systems provide an initial instruction or list of commands, some offer you an exit choice at every command. Some, however, expect you to not only know how to disconnect but when. One system provides an exit code to be issued at its standard prompt:_
At "--->" enter "EXIT"
The response is "Thank you for using our system."
And if you enter EXIT again it will thank you again. You've exited but still don't know where in the system you are.
Among the possible exit commands are:
EXIT BYE X QUIT
END STop ESC, ####
% LOGOFF UNDIAL QUIT
Even for vendor supplied systems logoff commands are not necessarily the same. If you are searching a NOTIS system, for instance, the LOGOFF varies:
QUIT or EXIT DELCAT, University of Delaware
LOGOFF LUIS, Florida State University Systems
<CTRL E>%QUIT MIRLYN, University of Michigan
OFF ELIXIR, SUNY Binghamton
<CTRL Z> Texas A&M Library
<KEYPAD ENTER, PF4> LUIS, University Texas Arlington
% LUIS, Northwestern University Library
In each case, however, this may only exit the library module and not the INTERNET connection. Minnesota's NOTIS system, according to Art St. George's "Internet-accessible Library Catalogs & Databases "...is considered by some to be relatively difficult to use, involving complex logoff escape sequences, for example. Be prepared to break the connection from your end, or just hang up, if necessary."
Hanging up the remote is more complicated than just disconnecting the call. Those systems that have no specific logoff command require that you terminate the TELNET connection. This could be relatively simple. A TELNET command line can be summoned in line mode by hitting a function key other than PF1, PF2 or PF3. A "?" will bring up a menu of choices; one of which is "QUIT." Therefore, in some systems, exiting is through a TELNET disconnect: PF4, for instance, then QUIT.
While access to multiple library catalogs is a tremendous benefit, there are additional advantages to the INTERNET. There are, of course, a variety of library-created databases and tools that have been established on local systems, although many are restricted to local IDs. While Dartmouth and CARL provide access to Grolier's encyclopedia, these are not available to remote users. On the other hand, Rensselaer offers an index to architectural slides and IEEE articles. Carnegie Mellon includes local bibliographies and indexes to architectural pictures, ICPSR data tapes, and has proposed an index to artificial intelligence literature. Brown University's Josiah offers expanded local indexes to its Special Collections.
Some of the databases available on the INTERNET are not library developed but are useful for special interests. These include the Latin America and legal databases at the University of New Mexico (connect with BOOTES.UNM.EDU and not LIBROS.UNM.EDU) and the agricultural and extension information available from the Pennsylvania Cooperative Extension. In addition, both Penn State University and the University of Maryland offer informational services about academic programs and activities at their campuses through an on-line system that can be called through the INTERNET.
Once adept at roaming the INTERNET, it is easy to branch out into additional byways of the national electronic system. BITNET is primarily an electronic mail service for communications. However, it has the capability to send the same message to thousands of users. As a result, on-line magazines and forums have been established. Of particular interest to librarians are the Public Access Computer Systems Forum, chaired by Charles W. Bailey, and the Newsletter on Serials Pricing Issues, edited by Marcia Tuttle. There is also a wide range of "magazines," "newsletters," and "informational exchanges" available electronically.
As useful as this broad range of tools might be, the barriers to access are clearly quite high. Many mainframe users will find the conventions and commands described here commonplace and can be almost eloquent about which operating system requires which obscure command. Many librarians will find all this daunting but will persevere because of the obvious benefits. Others, students and faculty, will not persist and will give up at the first blank screen or unintelligible prompt.
As our bibliographic tools become larger, more elaborate, more sophisticated, and more unfriendly, we will continue to distance the user from the information. The librarian must continue to mediate the search for knowledge. However, it should be remembered that most library users never fully mastered the card catalog either.
Return to Selected Papers and Presentationsinet.htm / August 2, 1995 / firstname.lastname@example.org