A previous version of this article first appeared in the Electronic Information column of the Optics and Photonics News, Sept/Oct 1993, under copyright of the Optical Society of America. The OPN editors have graciously allowed me to post it for general distribution, provided this statement remains with the article. _____________________________________________________________ The Internet by E-mail Jonathan Boswell is a physicist working for General Research Corporation, Vienna, Virginia. He can be reached at Internet address jboswell@grci.com Many of the articles in this column begin with a phrase of the form, "Anyone with access to the Internet can..." However, most small and many large corporations still do not provide full- service Internet connections for their employees. The subject of this month's column is about accessing the Internet by e-mail. Many readers, though lacking ftp (file transfer protocol) and telnet (remote login) access, may nevertheless have e-mail access to the Internet. Those who have AT&T Mail, MCIMail, SprintMail, America Online, GEnie, or CompuServe accounts, as well as BITNET and UUNET users, can locate Internet resources, perform file searches, and download free software via their e-mail gateways. A word of caution is required about costs. While those of us who enjoy a full Internet connection will regularly download megabyte files at no direct cost to ourselves, fees may be charged for both sending and receiving mail through Internet gateways. Check with your e-mail service provider for surcharges first, and then think twice about requesting large files. Sending Mail to the Internet First, you will need to figure out how to send mail through your gateway to the Internet. This information should be available from your e-mail service provider. A friend with access to the Internet can also get it for you by sending e-mail to mail-server@rtfm.mit.edu with a blank subject and a message body containing the command, send pub/usenet/alt.internet.services/Updated_Inter-Network_Mail_Guide An automated list server will send Scott Yanoff's Inter-network Mail Guide by return e-mail. A few examples from the Guide are reproduced below. Current Internet mail addresses are composed of a user name, followed by the "@" symbol, followed by a "domain" name. The "domain" name refers to a specific computer on the Internet, and it has a form such as "grci.com", or "stis.nsf.gov". Your e-mail gateway has the job of translating these character strings into the unique, numerical address of a specific computer on the Internet. Sometimes this translation can fail, which is why you may see two versions of an Internet address. One version uses the domain naming system, and the other uses the numerical address instead. In either case, the user name ahead of the "@" symbol corresponds to a specific mail recipient on that computer. For users of America Online and BITNET, the standard Internet addressing scheme, "user@domain", should work. On CompuServe, prefix the Internet address with the word, "Internet:". On AT&T Mail, send to "internet!domain!user". On MCIMail at the "To:" prompt, enter "Recipient (EMS)"; at the "EMS:" prompt, enter "INTERNET"; finally at the "Mbx:" prompt, enter the Internet address. Of course you do not actually type any of the quotation marks in the above examples. Now that you know how to send e-mail to the Internet, where should you send it? Fortunately, books called "resource guides", have been written to solve this problem. Resource Guides Many excellent resource guides exist. They are written by experts who have spent months or years exploring the Internet, discovering new file archives, databases, supercomputer research centers, and other neat things. My personal favorite is a relatively small (~75 page), well-maintained guide by the SURAnet Network Information Center of the University of Maryland. It is available as file pub/nic/infoguide.XX-XX.txt on ftp server ftp.sura.net, where XX-XX is the most recent date of publication. (See e-mail ftp example below.) You should be able to access all resources listed in this guide, with the exception of those requiring telnet (remote login). Recently the mother of all resource guides, a 200-page tome maintained by BBN Systems and Technologies (Cambridge, MA) for the National Science Foundation, was transferred to the InterNIC Directory and Database Services being provided by AT&T (South Plainfield, NJ). Specifically, you may send e-mail to mailserv@ds.internic.net, with the command help in the message body. The "help" command will cause the mail server to return its user's guide to you. If you want the last BBN (March 1993) resource guide, which is over 0.5MB, add to the message body the command: file /resource-guide/wholeguide.txt Unfortunately, this complete resource guide is no longer maintained as such. The new form is to spread the information out in text files on subdirectories of the ds.internic.net node. There is also no easy way to retrieve all of these files at once; they are stored in small chapter sections of approximately 1 page per file. You can have the full path names of all current resource guide sections sent to you with a command in the message body such as: send /resources/* Another mail message can then be sent to ds.internic.net with a "file" command for each desired section. Another InterNIC node is maintained for the NSF by General Atomics (San Diego, CA), and its mail server and file structure are much more user-friendly. Send mail to mailserv@is.internic.net with the two commands HELP INDEX in the message body. This will cause the server to return its user's guide, and a listing of its entire, intuitive directory structure and all files thereon. News Groups Knowing something sooner than the competition can give you the edge. Today, most academicians and government employees (in science and engineering disciplines) are linked by the immediacy of direct e-mail, as well as indirect e-mail provided by some automated mail exploder or moderated interest group. There are hundreds of these electronic conferences open and active on the Internet at any one time. (An electronic conference is a generic term including Usenet news groups, forums, bulletin boards, interest groups, discussion lists, E-newsletters, and E-journals.) Most of these news groups are open to anyone on the net at no cost, and you can electronically subscribe yourself if you know where to send your request. Care should be taken in coaxing broad-based "Lists of Lists" from the Internet, particularly if you are charged for inbound messages. You could easily find multi-megabyte files dumped in many separate messages in your "In" box. For example, SRI International (Menlo Park, CA) maintains a 1.3 MB file which attempts to capture all known Internet interest group mailing lists. If you want a copy of this list, send mail to mail-server@nisc.sri.com, with the single command, "send interest-groups" in the message body. A narrower, edited list is available from the Directory Team and Kent State University Library by sending mail to listserv@kentvm.kent.edu. Leave the subject and other info lines blank, and in the message body, type GET ACADLIST README GET ACADLIST FILE6 The first "GET" command will give you the user's guide and table of contents of the whole "ACADLIST", which is a list of only those e-conferences of academic interest. The second command should return the physical sciences list. If you find a news group you would like to join, send a subscription request to the server identified on the ACADLIST subscription line, (the SU: field), as follows. For LISTSERV, COMSERVE, MAILSERV, MAILBASE, NEWSERV, or SARASERV lists, leave the subject line blank and place in the message body the command SUBSCRIBE LISTNAME YourFirstName YourLastName YourInstitution where LISTNAME is the actual name of the news group you wish to join. Note that you subscribe to Internet interest groups by sending e-mail, not to the listname@domain address, but rather to listname-request@domain. This will get your message to the list coordinator, rather than the entire current membership of the list. Check the ACADLIST file's subscription line for the presence of the "-request". A mistake here could embarrass you before thousands of your colleagues. E-mail Archie The databases, technical papers, and software available from the Internet are stored in hundreds of computer archives all over the world. Many of these archives allow anyone (with full Internet access) to log in using the "anonymous" username, and download files for free. The problem is finding the data or programs that you want, among the millions of files on the hundreds of archives in service. Fortunately, there is archie, an archive listing service capable of searching its database of all filenames and directory names on all known (anonymous) archive sites. Jon Gilligan wrote about the e-mail interface to archie in this column last December. E-mail archie is briefly reviewed here. Send e-mail to one of the archie sites listed in Table 1. For a user's manual, send the word "HELP" somewhere in the subject or message body. The Rutgers archie server will send the lengthiest help file, including the archie manual pages. To perform a substring character search on program names, use the commands set search sub prog string where string is the substring you wish to search for. The "set search sub" command is necessary on some archie servers whose default operation requires an exact match. ________________________________________________________________ archie@archie.sura.net (University of Maryland) archie@archie.rutgers.edu (Rutgers University) archie@archie.unl.edu (U. of Nebraska, Lincoln) archie@archie.ans.net (ANS) archie@archie.au (Australian) archie@archie.funet.fi (Finland) archie@archie.doc.ic.ac.uk (UK/Europe) archie@archie.cs.huji.ac.il (Israel) archie@archie.wide.ad.jp (Japanese) archie@archie.ncu.edu.tw (Taiwanese) archie@archie.sogang.ac.kr (Korean) archie@archie.nz (New Zealand) archie@archie.kuis.kyoto-u.ac.jp (Japan) archie@archie.th-darmstadt.de (Germany) archie@archie.luth.se (Sweden) Table 1 - Archie Server Internet Domain Names ________________________________________________________________ As an example, we might like to find a BASIC version of a program called "uudecode." We could find one as shown in Figure 1. ________________________________________________________________ (send mail using appropriate e-mail gateway address) To: archie@sura.net Subj: set search sub prog uudecode Figure 1 - Example of E-mail Archie Search Request ________________________________________________________________ A greatly-edited portion of the search result is shown in Figure 2. ________________________________________________________________ Host ftp.bio.indiana.edu (129.79.224.25) Last updated 07:36 26 Apr 1993 Location: /util/ibmpc FILE -rw-r--r-- 1987 bytes uudecode.bas FILE -rw-r--r-- 3089 bytes uudecode.c Figure 2 - Edited Example of Archie Search Result ________________________________________________________________ E-mail ftp Once an interesting file or archive is located, it is possible to retrieve it using "anonymous ftp". Ftp is an acronym which stands for "File Transfer Protocol". But this is possible only if you have ftp access to the Internet. Fortunately, Digital Equipment Corporation (Maynard, MA) maintains a node and software which will accept specially-formulated e-mail commands, perform the anonymous ftp for you, and return the file by e-mail. For a brief user's manual on this service, send the word "help" in the message body to ftpmail@decwrl.dec.com. (BITNET users should obtain help and send ftpmail requests to BITFTP@PUCC. Some other ftp mail servers are listed in Table 2.) Most ftpmail commands are simply ftp commands for the server to execute on your behalf. You place these commands in the message body. The most useful of these commands are: connect serveraddress chdir directory dir chunksize bytelength get filename where serveraddress is the Internet domain name of the ftp archive server (identified by archie or other means), and directory is the full directory specification. Notice that there are at most one "chdir" command, and at most 10 "get" commands per mail message request. An example which retrieves the SURAnet resource guide (during the month of October) is shown in Figure 3. _________________________________________________________________ (send mail using appropriate e-mail gateway address) To: ftpmail@decwrl.dec.com Subj: connect ftp.sura.net chdir pub/nic dir chunksize 49000 get infoguide.10-93.txt Figure 3 - Example ftpmail Request _________________________________________________________________ This example will send approximately 150kB to your mailbox, in 49k chunks. The reason for the partitioning of large files has to do with limitations of some parts of the Internet, and limitations of the ftpmail server software. You can modify the chunk size up to 100kB with the "chunksize" command. The partitioning of large files may seem awkward, but it gets even worse. You can also have the ftpmail server send binary files back to you. This is done using the following commands, before the "get" command. binary uuencode The "binary" command will cause ftpmail to get the file in binary mode. The resulting (temporary) file on the ftpmail server will be unmailable. The "uuencode" command causes each byte of the binary file to be encoded in plain text characters. The encoded file is then mailed back to you. Obviously you have to reverse this process in order to recover the original binary file. The program you will need to decode a uuencoded mail file is called "uudecode". (Patch the partitions back together again first, being sure to edit out all mail headers/footers.) The e-mail archie example in Figure 2 lists the location of a BASIC version of this program, which is itself mailable as text. Macintosh owners have it both simpler and harder. Many archives featuring Macintosh software and data have already "hexified" the binary file. These files can be identified by the ".hqx" extention, and are directly mailable without the "uuencode" command. Unfortunately, there is no common compiler sitting on everyone's Macintosh with which to build a dehexification program. You will have to locate a program, such as BinHex, on a direct-dial server, on-line service, through a friend, or by purchasing it. (One such supplier can be reached at 805-484-9400). ________________________________________________________________ ftpmail@decwrl.dec.com (US) BITFTP@PUCC.Princeton.edu (US) ftpmail@grasp.insa-lyon.fr (France) BITFTP@vm.gmd.de (Germany) Table 2 - ftp Mail Servers. These machines have inconsistent commands. For help on command syntax, send a message to the target machine with the word "HELP" in the message body. ________________________________________________________________ Famous Last Words The above commands were tried successfully over CompuServe(1), MCIMail, and America Online. If any of them do not work, check first to make sure that the correct target machine actually accepted your message. Many addressing errors on the Internet result in "bounced mail", wherein your mail message is sent back to you with an error, such as "User unknown". If the target machine accepted the mail message but generated errors, check to make sure that your mail message included commands in the correct case. Most nodes on the Internet are Unix machines, and nearly all commands and filenames are case sensitive. If your corrected commands still do not work, it is possible that filenames and services have changed in the months since this column was written. The Internet is a rapidly growing, dynamic network without any centralized control. Don't be afraid to experiment with archie, or follow up on new archives identified in a resource guide. Reliance on your e-mail interface will slow you down a bit. And some database services require remote login capability. But much of the information, and all of the people, on well over a million computers are accessible to you, and they're only a mail message away. Notes: 1) CompuServe members will have difficulty with "text" files containing 8-bit characters, when using ftpmail. If you have trouble getting all parts of files through the CompuServe mail gateway, you may either add the "binary" and "uuencode" qualifiers to your ftpmail commands, or employ BITFTP@PUCC.Princeton.edu. BITFTP correctly zeros the high-order bit of 8-bit characters. ftpmail does not.