March 1996 Date: Mon, 4 Mar 96 00:16:52 UT From: "Loren Toomsen" Subject: RE: Littlefield links ---------- From: Bill Minnick Sent: Sunday, February 04, 1996 12:33 AM To: Greg & Andrea Littlefield Cc: ann.turner@239.microbbs.us.com; CJHARLESS@aol.com; MRS CAROL R AUSTIN; Claiborne Duval; Dr. Michael Austin; eph1@ix.netcom.com; fnordin@ix.netcom.com; Gary Alan Parsons; genweb@ucsd.edu; Jan Unter; jcarlin@pacificnet.net; Loren Toomsen; wwfamily@xmission.com; gathman@lava.net; Timothy B. Doyle; al.buskirk@239.microbbs.us.com; Alice Martin Subject: Re: Littlefield links >Hi, I just wanted to let you all know that I added links from the >Littlefields in your database to their descendants in mine. I will try >to add reverse links (from mine to yours) asap, but for the time being, >I have a link to 'our Austin cousins' on my genealogy page. Thanks for >your time! > >Greg Littlefield >Austin, TX >http://www.eden.com/~gregandi/genealogy.html GREG: This is exactly what I envisioned 10 months ago when I colaborated with Tim Doyle of Dallas, TX to add the "LINK" insertion capability in our Austin and other genealogy pages. Thanks for seizing the opportunity to add to our Richard Austin database by linking in your Littlefield database. This is the next level of the GENWEB concept at work. For those who may not see the impact of this event, Greg has, on his own, and without our intervention, added a link to his Littlefield database into our Richard Austin Database via the automatic LINK feature on every page of Tim Doyle's genealogy databases. With the LINK feature, the simple action of one person, Greg, adds real value for all other researchers to both the Austin and Littlefield databases. This example is a significant new application of the Web that uses the potential of the Web to take publication of genealogy beyond what is possible in the printed book. Thanks, Greg, for following our lead in this evolutionary step in development of the GenWeb concept. Regards, Bill Minnick Austin Web Project NOTE: The Link inserted by Greg can be seeen at URL: http://sillyg.doit.com/cgi-bin/igmget/n=Austin?91907 To: genweb@UCSD.EDU From: Walter Subject: genealogy seeking surname KLEINER originally from Praha Czechoslvakia can provide a little more information with respect to their approximate births and deaths Date: Mon, 4 Mar 1996 15:13:36 +0100 From: BOUVIER Regis CNET ISSY (Tel 45296729) To: GENWr (Non Receipt Notification Requested) Je cherche de l'information sur la famille Degueldre Emile Degueldre a epouse Augustine Wuillame a` Givry (Belgique) vers1890 I'm looking for information on the family Degueldre Emile Degueldre married Augustine Wuillame 1890 in Givry (Belgique) From: "W. Wesley Groleau (Wes)" Subject: An odd GEDCOM/HTML idea To: genweb@UCSD.EDU, gedcom-l@vm1.nodak.edu Date: Thu, 7 Mar 96 11:48:23 EST In-Reply-To: <4665.tfj@apusapus.demon.co.uk>; from "Trevor Jenkins" at Nov 9, 95 11:02 pm Mailer: Elm [revision: 70.85] I know there are better ways to publish data, but this one is so EASY that someone might want to do it just for a diversion: Transform a GEDCOM file by merely placing 
 (
) before (after) turn all cross-ref markers into NAME anchors turn all cross-ref pointers into HREF anchors back to the NAME. So 0 @any_string_at_all@ INDI 1 NAME John Doe 1 FAMC @another_string@ becomes 0 @any_string_at_all@ INDI 1 NAME John Doe 1 FAMC @another_string@ The GEDCOM is displayed with NO visual transformation other than whatever the browser does to highlight a link. And navigation is easier than with a plain text file. No information is lost by the inability of a GEDCOM-to-HTML converter to handle Acme's GEDCOM format or X-Mart's custom tags. Links to such tags in other people's GEDCOMs would be simple (though volatile): just cut and paste their anchor, change NAME=" to HREF="URL# The converter would not mess these up because they would not have the @ signs.gn A program for this would be trivial, and would even work on "illegal" GEDCOM. If I wasn't so hardware lame at the moment, I'd do it myself. I just might anyway. But even so, I do agree that Gene Stark, Scott McGee, and others are generating BETTER formats than the above. --------------------------------------------------------------------------- W. Wesley Groleau (Wes) Office: 219-429-4923 Magnavox - Mail Stop 10-40 Home: 219-471-7206 1010 Production Road QuickMail (Mac): wwgrol@most.fw.hac.com Fort Wayne, IN 46808 elm (Unix): wwgrol@pseserv3.fw.hac.com --------------------------------------------------------------------------- Saturday, March 9, 1996 8:36:04 PM GenWeb Item From: ghoffman@UCSD.EDU,Internet Subject: We've Been Spammed! To: GenWeb To GenWeb mailing list subscribers, Yes, you did receive a notice through this mailing list about a Hong Kong film playing on the campus of UC San Diego next Wednesday. Every public mailing list maintained by UCSD was spammed by this announcement for the Tai-Chi Master film at the student center. Long-time net users are used to a certain amount of "noise" in the channel. However, you can be sure the perpetrators of this prank will hear get an earful from the mailing list managers. Please do not compound the problem by replying to the noise messages in the genweb mailing list. Usually, this just causes the annoying message to be repeated to all subscribers again -- even you. And please stick with us as we work this out with the spammers; don't unsubscribe at the first annoying message. And please, if you would like to unsubscribe, please send that message to listserv@ucsd.edu, not to this list. Cheers, Gary -- *************************************************************************** *Gary B. Hoffman, Computing Services Manager e-mail: ghoffman@ucsd.edu* *Graduate School of International Relations and Pacific Studies (IR/PS)* *University of California, San Diego (UCSD) voice: (619) 534-1989* *9500 Gilman Dr., La Jolla, CA 92093-0519 USA fax: (619) 534-3939* *************************************************************************** From: KGraves239@aol.com Received: by emout06.mail.aol.com (8.6.12/8.6.12) id LAA05498 for genweb@ucsd.edu; Mon, 11 Mar 1996 11:28:07 -0500 Date: Mon, 11 Mar 1996 11:28:07 -0500 Message-ID: <960311112805_347808215@emout06.mail.aol.com> To: genweb@UCSD.EDU Subject: Conversion of Genealogical Data from Text & WP First a question/comment. As I have been monitoring this list for the past several months, I have noticed a large variation in number of messages. For a while around the end of 1995 I was deluged with meaningful discussion. Then nothing for a long period of time. Are these breaks real or is there discussion going on somewhere else that I don't know about? My real question is regarding my involvement with a one-name society (Graves Family Association) and data gathering program for the Graves/Greaves surname and variations. Since much information and all older genealogies have been compiled and published either as non-computerized text (in books, etc.), or in word processing format, we have a need to convert from either of those formats to a database and then to gedcom,etc. Since scanning plus OCR will get printed text into word processing format, the real question is going from word processing to database and gedcom. I have written programs to do some pieces of this, so I know it can be done. It is also apparent that conversion can go from gedcom to various standard genealogical publication format, since many programs already do that. Has anyone already done the conversion from text or word processing format to database, gedcom, etc.? What do you think is the best way to do it? Any other comments? Ken Graves From: JohnR238@aol.com Received: by emout05.mail.aol.com (8.6.12/8.6.12) id VAA29140 for genweb@ucsd.edu; Mon, 11 Mar 1996 21:17:17 -0500 Date: Mon, 11 Mar 1996 21:17:17 -0500 Message-ID: <960311211716_166196923@emout05.mail.aol.com> To: KGraves239@aol.com cc: genweb@UCSD.EDU Subject: Re: Conversion of Genealogical Data from Text & WP In a message dated 96-03-11 11:47:40 EST, you write: >Has anyone already done the conversion from text or word processing format to >database, gedcom, etc.? What do you think is the best way to do it? Any other >comments? > >Ken Graves My thoughts are that existing text files are too unstructured to write a program to convert to a data base. I would probably extract every name into a gedcom compatible program, then chop up the text into comments files (in whatever way the particular program wants to see them). That way you've got a beginning file that can "travel" as a gedcom, and you can work on particular people's records as time permits. This procedure has worked well for me - of course you must still OCR the books as a first step, then if the data is structured enough, or has an index, you may be able to machine extract the names, but I've generally just read through the text, and added the names only to my database program as a first pass. John Rigdon Web Index Master To: KGraves239@aol.com, genweb@UCSD.EDU From: ghoffman@UCSD.EDU (Gary Hoffman) Organization: IR/PS UC San Diego, La Jolla CA 92093-0519 Date: Tue, 12 Mar 1996 10:33:16 PDT Subject: Re: Conversion of Genealogical Data from Text & WP To GenWebbers: Ken Graves asked: First a question/comment. As I have been monitoring this list for the past several months, I have noticed a large variation in number of messages. For a while around the end of 1995 I was deluged with meaningful discussion. Then nothing for a long period of time. Are these breaks real or is there discussion going on somewhere else that I don't know about? Ken, these are real breaks, taken to allow us to catch our wind. This GenWeb stuff is pretty heady and we occasionally get hyperventilated. The messages to this list are archived at URL http://demo.genweb.org/genweblist/ and show the variability in message traffic. Thanks for sticking with us during the lean periods. Get ready for another surge. Cheers, Gary *************************************************************************** *Gary B. Hoffman, Computing Services Manager e-mail: ghoffman@ucsd.edu* *Graduate School of International Relations and Pacific Studies (IR/PS)* *University of California, San Diego (UCSD) voice: (619) 534-1989* *9500 Gilman Dr., La Jolla, CA 92093-0519 USA fax: (619) 534-3939* *************************************************************************** From: JohnR238@aol.com Received: by emout07.mail.aol.com (8.6.12/8.6.12) id WAA20842 for genweb@ucsd.edu; Tue, 12 Mar 1996 22:52:50 -0500 Date: Tue, 12 Mar 1996 22:52:50 -0500 Message-ID: <960312225248_445075687@emout07.mail.aol.com> To: KGraves239@aol.com cc: genweb@UCSD.EDU Subject: Re: Conversion of Genealogical Data from Text & WP Thanks for your reply. I appreciate your frustration with the Gen Web list, but I am forwarding this reply to the list in case some others want to join the discussion. First to the mail issue. I think it's some more of AOL's gremlins showing up again. I archive everthing I can get my hands on related to genealogy and it's really a pain at times. Currently I have switched ALL my major list subscriptions (roots, etc) to a local provider because of the problems with AOL. I think part of the problem we've seen lately with gen-web is also caused by AOL. but hey - when you dance with an elephant, you take your chances. I have never found a commercial or shareware program for the pc that does a really creditable job of indexing text. The only one which came anywhere close is called PC-Index and it is no longer supported by the vendor. It was written in Turbo Pascal in the early 80's so it could use a face lift, but it does kinda work. Unfortunately it's an all word indexer, which does nothing for first and last name associations and it also makes a mess of dates. I have taken upon myself a project for building a genealogical index of the web, indexing the names found in files, not just the sites and file names that we now have. I use a collection of routines I've written and often write new routines on the fly, but I know I'm still not getting even a major part of the names in totally unformatted text. I have a program called AutoIndex which is on the web at http://home.aol.com/JohnR238 that I have written and use on these files. Your comments and feedback are appreciated, but I think I've pushed the program about as far as it can go, and I am currently rewriting it for Windows. I'm thinking now that some kind of interactive drag and drop program will be needed to speed up the manual process, but I can't develop the logic to fully extract and index text files appropriately. One prime example of the problem is found in the following excerpt: She died in 1838. James A. Rigdon received his education in the primitive manner of former times and was but a small child when his father died. He made his home with his uncle a number of years, and in 1835 married Miss Josephine Kirchner, a native of Rhine, Germany. When she was about twelve years old her parents came to America and settled in Ste. Genevieve County. They had 15 children: Mary (V. Gesey), Lewis, Ellen M (William Rudluff), Emily B. (John Harter), John L., Joseph, Peter, Mary Ann (Tobias Thomure), Felix J., Rosale (Joseph Gery), Lucinda (Walter Sharlock), Francis A., Michael B. and two deceased. My passage of this paragraph through Auto Index yielded only: America ~COUNTY, Ste Genevieve 001-1 COUNTY, Ste Genevieve ~ 001-1 GERMANY, Rhine ~ 001-1 KIRCHNER, Miss Josephine ~ 001-1 RIGDON, James A ~ 001-1 >From my perspective it got one totally bogus entry in the "America..." entry and totally ignored the 15 kids names which had first and sometimes married names only. I just don't think there's any way to write the logic to deal with such a mess. John Rigdon Web Index Master From: JohnR238@aol.com Received: by mail06.mail.aol.com (8.6.12/8.6.12) id GAA00558; Thu, 14 Mar 1996 06:11:37 -0500 Date: Thu, 14 Mar 1996 06:11:37 -0500 Message-ID: <960314061136_350852276@mail06.mail.aol.com> To: jack@cyberstation.net cc: genweb@UCSD.EDU Subject: Re: Conversion of Genealogical Data from Text & WP Thanks for your interest, and perhaps working together, we can move this idea forward, but I'm resigned to the fact that a listing such as the one below which we find all the time in genealogy works cannot be machine extracted reliably. There's just no to put a first and last name together through computer logic rules. >> Lewis, Ellen M (William Rudluff), Emily B. (John Harter), John L., Joseph, >> Peter, Mary Ann (Tobias Thomure), Felix J., Rosale (Joseph Gery), Lucinda >> (Walter Sharlock), Francis A., Michael B. and two deceased. >> Lewis - should be RIGDON, Lewis - and if I didn't know better, I would index it as LEWIS, Ellen M. which produces a totally bogus entry Ellen M - Should be RUDLUFF, Ellen M Rigdon William Rudluff - Should be RUDLUFF, William Emily B - Should be HARTER, Emily B. Rigdon John Harter - Should be HARTER, John John L. - Should be RIGDON, John L. While there's definately a pattern, how do you write that - and while I may write a routine for a particular text file, chances are it won't apply at all in the next book - or even in the next paragraph of the same book. Any ideas are appreciated. From: JohnR238@aol.com Received: by emout05.mail.aol.com (8.6.12/8.6.12) id SAA11893; Thu, 14 Mar 1996 18:33:47 -0500 Date: Thu, 14 Mar 1996 18:33:47 -0500 Message-ID: <960314183346_446605048@emout05.mail.aol.com> To: wwgrol@pseserv3.fw.hac.com cc: genweb@UCSD.EDU Subject: Re: Conversion of Genealogical Data from Text & WP In a message dated 96-03-14 08:43:13 EST, you write: >Another option: Suppose you have a plain-text file, AND a GEDCOM that >has some overlap with it. It would not be very difficult to tell it to > >FOR each NAME in the GEDCOM LOOP > find every occurrence of that name in the text and turn it into > an index entry >END LOOP > >(Of course, you usually DON'T have the GEDCOM) > >-- >---------------------------------------------------- This is basically the indexing option that Word Perfect offers now. The overall process of indexing is easy enough, but it is a manual process. If we were indexing for surnames only, we could probably develop a master list which would accomplish 90 - 98% success fairly easily. And maybe this is the "realistic" type of index to build, rather than an all name index. >From my perspective, this kind of index for the top 2,000 surnames would be useless because of the number of names it'd ultimately have. On the other hand, I'm interested in RIGDONS anytime, anywhere and anytime I see the name, I feel like I've found a golden nugget, regardless of how remote the reference. Maybe we need a project to find the bottom 98,000 surnames anytime anywhere and for the top 2,000 try to index both first and last names. I know for myself that I now have over 2,000 sites with genealogy materials which I have not visited or indexed and the couple of hundred sites I have indexed has yielded 2,000,000 names. I can only index 10 - 15 sites per week, and my unvisited sites list is growing at the rate of 50 - 60 per week. Methinks I'm facing a losing battle here. John Rigdon Web Index Master From: JohnR238@aol.com Received: by mail06.mail.aol.com (8.6.12/8.6.12) id MAA26678; Sat, 16 Mar 1996 12:47:29 -0500 Date: Sat, 16 Mar 1996 12:47:29 -0500 Message-ID: <960316124726_447931807@mail06> To: GEM@laspau.mhs.harvard.edu cc: genweb@UCSD.EDU Subject: Re: Conversion of Genealogical In a message dated 96-03-16 00:58:17 EST, you write: >What kind of hardware and software are you using to do the indexing? Is >it something that is suitable to be done by multiple people? How much of >your time does it take? It seems plausible that a few more people >working on the immediate problem of indexing could up the rate to a point >where we could keep up I've written the various programs myself, and will be happy to let you have a copy if you want to help. Obviously, any help we/I can get on this project is going to help the index - and the users along. I'll need to document the usage of some of the routines and they're rather quick and dirty - DOS prompt kinda things with no real user interface. Many of the routines I write/have written on the fly to handle particular file formats I've found. Their accuracy in extracting names ranges from 100% on GEDCOM and other "structured" lists to about 25% on totally unstructured, text data. I see two areas that I could use some volunteer help and I'm going to start pushing the idea on my web page since I'm now gitting 200 - 300 downloads a day of the files. 1. Take the routines - or use your own and index sites which have not yet been visited. 2. Manually update/correct the indexes for sites which are text only. This will take more time and effort on someone's part, but obviously the IMPORTANT name is the one we missed on the first pass. For folks who want to contribute some time in this area, we can xfer the file via email, and they can create the index/update it at their leisure, then xfer the index in for merging with the master index. I'm spending about 4 - 5 hours per week now - which typically extracts 20,000 - 30,000 names. If I do a run on GEDCOM files as I've done this past week, I get much higher numbers (225,000 names for the same amount of time and less effort). Others have / are undertaking to index GEDCOM files and I don't see my effort conflicting with theirs, but rather complementing. My goal is to provide a "lowest level" all name index to as many sites and names as possible. With a name as obscure as mine, I'm always scraping for RIGDONS anytime and anywhere. John Rigdon Web Index Master Date: Sun, 17 Mar 1996 12:47:55 -0800 From: macdonalda@wvin.com (MacDonald Andera) X-Mailer: Mozilla 2.01 (Win95; I) MIME-Version: 1.0 To: genweb@UCSD.EDU Subject: Ukraine Research Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit > I have been approached by a new member of my genealogical society in the hopes that I > can help him locate people involved in Ukraine Research. If anyone out there has > knowledge of any society, club, group or person(s) having specific knowledge in this > area, please send me their e-mail or Postal address so that I can put this young man in > touch with them. Research in the area he is from is very difficult at this time as the > border between Poland and Ukraine keeps shifting. Records could be in many places. > This young gentleman is just beginning his quest and it would be wonderful if we could > provide him with some basic information to whet his appetite. Thank you. To: genweb@UCSD.EDU From: ghoffman@UCSD.EDU (Gary Hoffman) Organization: IR/PS UC San Diego, La Jolla CA 92093-0519 Date: Mon, 18 Mar 1996 21:36:55 PDT Subject: You are tuned to GenWeb - Do not reply Anyone receiving this message is one of nearly 600 people subscribed to the e-mail mailing list GENWEB. The purpose of this list is to facilitate the development of a linked, worldwide distributed genealogy database. If this topic is not of interest to you ... here is how to unsubscribe: Send an e-mail message to listserv@ucsd.edu In the body of the message put the words: UNSUB GENWEB That's all. Do not reply to this message. Do not send these commands to genweb@ucsd.edu. Do not send me a message about unsubscribing. Just do it as outlined above. If you still want to read about the GenWeb, please point your WWW browser to the URL http://demo.genweb.org/genweblist/genweblist.html All current and archived messages are there for your perusal without cluttering your mailbox. Thanks, Gary *************************************************************************** *Gary B. Hoffman, Computing Services Manager e-mail: ghoffman@ucsd.edu* *Graduate School of International Relations and Pacific Studies (IR/PS)* *University of California, San Diego (UCSD) voice: (619) 534-1989* *9500 Gilman Dr., La Jolla, CA 92093-0519 USA fax: (619) 534-3939* *************************************************************************** From: Cliff Manis Message-Id: <199603190847.RAA12138@soback.kornet.nm.kr> Subject: GenServ: 3 Million Names GEDCOM databases To: genweb@UCSD.EDU Date: Tue, 19 Mar 1996 17:47:16 +0900 (KST) Cc: gedcom-l@vm1.nodak.edu, lines-l@vm1.nodak.edu, cmanis@soback.kornet.nm.kr (Cliff Manis) X-Mailer: ELM [version 2.4 PL21-h4] MIME-Version: 1.0 Content-Type: text/plain; charset=iso-2022-kr Content-Transfer-Encoding: 7bit GenServ: Genealogical GEDCOM Server System On-Line Internet The main objective of the GenServ system is to collect data and make that data available to others as On-Line research material. Collecting data and sending out messages. It could be very important in helping you to find DIRECT Family information. We do have data/users here from the US, Canada and more than 27 countries. We now have over 3,000,000 names on this system for the users to use as research material. Maybe the GenServ is a place to find some more information about your direct ancestors and maybe we have them now in the more than 2000 different GEDCOM databases on this system. We have more than 148,000 different surnames available now. Please visit our GenServ homepage and take a look. URL below. http://soback.kornet.nm.kr/~cmanis/ Please note that we collect GEDCOM data on this system. When someone sends us a GEDCOM file, we give them access to all the other data on this system. We do not accept GEDCOM datafiles via email, so please read the information below. We try to keep it up to date with the latest happenings of news in the Genealogical world, and have points from the GenServ homepage which I feel are important. I have been working my own genealogy since 1955 and collecting family data. I hope you find the GenServ system homepage interesting and helpful. We will try to keep it updated and available for you to find new information. Search for "GenServ" on any of the major SEARCH utilities on the WEB and you will find us. Visit our homepage and request the GenServ Documentation to find out the answer to questions below and many more answers to your questions.. How you can be a user of the GenServ system ? Send me a gedcom data file via postal mail. How much do we charge ? Only one Dollar a Month ! $12 a year. Who can have access to the data on tis system. Only the User of the GenServ system. The ones who send data. You want information about a surname ? This system will allow ONE FREE Surname Search for any one name. Visit our homepage. Most people are looking for what, when, where and how information. The GenServ Documentation has all these answers I started this GenServ System in 1991, so it's not just a Jack-in-the-Box that started yesterday. Thanks for your interest. Don't send me a note, visit the homepage or request the DOCs below. It has all the answers. The latest GenServ DOCs dated 1 Jan 96 may be obtained anytime by just sending any message (yes even a blank one) to genserv-doc@ProgCons.COM That mailbox is dedicated to sending out the GenServ Documentation. Send it a message today. Request the DOCs the easy way. Best of luck to you and your research in 1996. Cliff -- Cliff Manis cmanis@progcons.com // cmanis@soback.KorNet.nm.kr Seoul, Korea GenServ "Genealogical Server" a service for making GEDCOM data available. OVER --> 3,000,000 names in this GEDCOM database server WWW Genserv Homepage URL: http://soback.kornet.nm.kr/~cmanis/ - Date: Thu, 21 Mar 1996 19:35:23 -0800 From: macdonalda@wvin.com (MacDonald Andera) X-Mailer: Mozilla 2.01 (Win95; I) MIME-Version: 1.0 To: genweb@UCSD.EDU Subject: [Fwd: Ukraine Research] Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Andrea D. MacDonald wrote: > > > I have been approached by a new member of my genealogical society in the hopes that I > > can help him locate people involved in Ukraine Research. If anyone out there has > > knowledge of any society, club, group or person(s) having specific knowledge in this > > area, please send me their e-mail or Postal address so that I can put this young man in > > touch with them. Research in the area he is from is very difficult at this time as the > > border between Poland and Ukraine keeps shifting. Records could be in many places. > > This young gentleman is just beginning his quest and it would be wonderful if we could > > provide him with some basic information to whet his appetite. Thank you. Reply to macdonalda@misery.wvin.com Date: Thu, 21 Mar 1996 22:11:20 -0800 (PST) From: JAMES F SHAFFER To: Genealogy User Group Subject: METTAM Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Is anyone out there doing research on the METTAM surname. If so, please contact me at either this address or: James.Shaffer@sprint.sprint.com Thanks! From: "W. Wesley Groleau (Wes)" Subject: Stale links between databases. To: genweb@UCSD.EDU Date: Fri, 22 Mar 96 12:56:49 EST Mailer: Elm [revision: 70.85] know this is a recurring topic, and my idea may have already come up and been torpedoed, but here goes: How about genweb folks agreeing on a format (actually, it's pretty much already defined by CGI) for turning a fragment of GEDCOM into a string of CGI parameters. A CGI script that receives it would treat it as a search request: "Show me the equivalent portion of your database." Each database owner could define for himself how to respond, if at all. If someone successfully devises good heuristics, that can find a match without being nit-picky on unimportant details (since the same INDI can be expressed many ways in GEDCOM) this could be used as a link that is less volatile than an INDI number or a person's name. Note that I did not say _non-volatile_, I said _less_volatile_ GenServ, GIM, and existing LifeLines report programs are possible starting points for matching algorithms. Since CGI limits length of a request, perhaps an alternative would be to send them a URL for your GEDCOM and a locator for the item you want to match. Tentative example: 0 @oiuyt@ INDI 1 NAME William Henry /French/ 1 NOTE >2 CONC Same person in Virginia Isle's database Hi, Wes. I made a similar suggestion some time ago. First, I believe that the only way to make this work is to carry the information that there *is* a link in the notes. Some programs have no provision for external labels. The one thing all generated gedcoms have in common is the notes. There is no concensus on this point. Second, others have suggested that in order to keep from having "stale links between databases" that there must be some "External File" which contains the pointer. This means that, rather than have each person have to change all their notes or whatever (see First, above) to point to the new location, they would only have to change the "External File" [EF]. Third, I suggested that the EF be maintained in the same way that Fidonet handles the maintenance of its nodelist. My proposal would be that each database be assigned a name. Mine could be jmurphy, for example. Let's also propose that yours is wgroleau, for the sake of argument. Now, assuming a link between an individual in my data and one in yours, I would add the note LINK wgroleau under that individual in my database. If you wanted to point to mine, you would say LINK jmurphy. The EF would look something like this: jmurphy http://genealogy.org/~smcgee/cgi-bin/genweb.cgi/DB=murphy/ INDEX=I[xxx]/?PedigreeInternal" wgroleau http://host/cgi-bin/GEDCOM-Match&http://genealogy.org/~wgroleau/WGROLEAU.GED &INDI+@oiuyt@ I can't solve the word-wrap problem here, but you get the general drift. And I'm not too sure how to specify a name to get to my database. (Scott, what would the code be if I wanted to pass you a particular name?) Now, the key to the solution of the problem is to agree on the steps. 1. We need to identify how the information that there *is* a link is taken from the genealogy software and passed to the gedcom. My proposal is that it be done in the notes. 2. We need to create a file of links, the EF. a. We need to define how the file is to be maintained and modified. I think it should be done by those where the data resides. b. We need to agree that the format includes two fields: a database descriptor (like wgroleau) and the link information. I omitted the because that can be handled by the html generator. c. We need to define how the updated file is distributed to those who are creating the html stuff. 3. Each html generator needs to be modified to search the notes for the LINK, hit the EF for the correct current link, and insert it into their generated page. 4. We need to discuss what should happen if a link occurs to a database that is not defined in the EF. My personal feeling is that it should just be ignored in that case. Can I have an "amen"? A "Whoa-Bundy"? Jeff Murphy Redmond, Oregon Muhlenberg Co., KY genealogy: http://www.teleport.com/~jmurphy/ subscribe to PAFHELP-L at majordomo@teleport.com To: genweb@UCSD.EDU From: ghoffman@UCSD.EDU (Gary Hoffman) Organization: IR/PS UC San Diego, La Jolla CA 92093-0519 Date: Sat, 23 Mar 1996 08:09:50 PDT Subject: Re: Stale links between databases Jeff Murphy wrote, in part: ---------------- Third, I suggested that the EF be maintained in the same way that Fidonet handles the maintenance of its nodelist. My proposal would be that each database be assigned a name. Mine could be jmurphy, for example. Let's also propose that yours is wgroleau, for the sake of argument. Now, assuming a link between an individual in my data and one in yours, I would add the note LINK wgroleau under that individual in my database. If you wanted to point to mine, you would say LINK jmurphy. ----------------- Let's not forget that we have the genweb.org domain to use for this purpose. I set this up about 15 months ago and even established some subdomains. This way, we can let the Domain Name System maintain at least the host part of the database for us. For example: murphy.surnames.genweb.org can resolve to whatever the name of your server is. All you keep track of is "murphy" and then plug it in to the rest of the domain name. Of course, there is the problem of various paths within each server. However, I believe dedicated genweb sites (or well formed ones) can have a universal path defined, such as ~genweb/xxxx. Cheers, Gary *************************************************************************** *Gary B. Hoffman, Computing Services Manager e-mail: ghoffman@ucsd.edu* *Graduate School of International Relations and Pacific Studies (IR/PS)* *University of California, San Diego (UCSD) voice: (619) 534-1989* *9500 Gilman Dr., La Jolla, CA 92093-0519 USA fax: (619) 534-3939* *************************************************************************** Date: Sat, 23 Mar 1996 11:35:49 -0500 Message-ID: <960323113548_175679140@mail06> To: jmurphy@teleport.com cc: genweb@UCSD.EDU Subject: Re: Stale links between databases. Jeff Murphy's approach to the index links makes sense and fits well with the project I am doing in the Genealogists Index to the World Wide Web. http://home.aol.com/genwebindx While my project encompasses not just GEDCOMS, but also text files on the WEB, my reference.txt could be easily modified to accomodate a potential "EF" format. In fact, the data elements I am keeping in addition to the URL in the reference file may be useful in the "EF" format. They are: date of last update, file size, and short description. This makes the "EF" file much more useful for hoomans, not just c'puters. John Rigdon WEB Index Master Date: Wed, 23 Mar 1994 12:28:14 -0800 To: ghoffman@UCSD.EDU (Gary Hoffman) From: Jeff Murphy Subject: Re: Stale links between databases Cc: genweb@UCSD.EDU At 08:09 AM 3/23/96 PDT, you wrote: >My proposal would be that each database be assigned a name. Mine could be >jmurphy, for example. Let's also propose that yours is wgroleau, for the >Let's not forget that we have the genweb.org domain to use for this >purpose. I set this up about 15 months ago and even established some >subdomains. This way, we can let the Domain Name System maintain at least >the host part of the database for us. For example: > > murphy.surnames.genweb.org > >can resolve to whatever the name of your server is. All you keep track of >is "murphy" and then plug it in to the rest of the domain name. Of course, >there is the problem of various paths within each server. However, I >believe dedicated genweb sites (or well formed ones) can have a universal >path defined, such as ~genweb/xxxx. Hi, Gary. I didn't realize you had created the domain. However, I don't really understand how that would work. Would each database have to carry the domain for each individual? Or do you see it taking the place of the external file I was talking about? I am unfamiliar with how domains work in the Internet, and suspect I'm not alone. I'd appreciate an explanation in a little more detail, for us novices. Are you saying that, properly maintained, the LINK would point to the domain, but the actual location of the database would be able to change whenever, without anyone having to go into their notes and modify the LINK command? Jeff Murphy Redmond, Oregon Muhlenberg Co., KY genealogy: http://www.teleport.com/~jmurphy/ subscribe to PAFHELP-L at majordomo@teleport.com Date: Sun, 24 Mar 1996 13:25:28 -0800 To: genweb@UCSD.EDU From: Jeff Murphy Subject: Sorting indexes by locality [Gene, this is mostly to you, I guess, but there are others who it might affect, and I wanted to hit everyone at once.] There was some discussion awhile ago about the possible ability to create an index sorted by location. The need for such a sort has now arisen, and I wanted to discuss it to see if it was possible. I have proposed to the KY researchers that we create a master web page that will take us to the various counties in KY, so that we can look for individuals by location. I am in the process of discussing the options still, but it looks very likely that we can set it all up. Many of us KY researchers have relatives in multiple counties. For instance, while I specialize in Muhlenberg, I have people in the 8 surrounding counties, and some elsewhere. But many people would not search a Muhlenberg database for people from Ohio, Butler, Trigg or other counties. It doesn't occur to them. In setting up a locality index for all names of those in KY (birth, death, burial, marriage), we would enable all those county pages to point to a single index, which would include all the current genweb sites. And yet it would be a subset, as it would not include anything not in KY. We would hope that as new genweb sites and databases are added they would be automatically included in an updated index - much like what Gene is doing now with his comprehensive index. I think it would be best if this were sorted by name within state, and showing the location, so if there are 5 John SMITHs, say, the user could go first to the one from the county nearest their research area. Your responses would be appreciated. Jeff Murphy Redmond, Oregon Muhlenberg Co., KY genealogy: http://www.teleport.com/~jmurphy/ subscribe to PAFHELP-L at majordomo@teleport.com To: genweb@UCSD.EDU From: ghoffman@UCSD.EDU (Gary Hoffman) Organization: IR/PS UC San Diego, La Jolla CA 92093-0519 Date: Sun, 24 Mar 1996 21:04:53 PDT Subject: How to do Virtual Hosting? To Genwebbers: The local Internet Service Providers in this area are advertising "virtual hosting" at their sites. That is, they will register a domain name such as "company.com" and then put a "www." in front of that to make a virtual web site using their host computer. This means that a browser call to "www.company.com" seemingly gets the index.html page at the root of the Web server directory tree. If there is more than one virtual web site on that server, there must be some mechanism to sort out the calls. So, a call to www.company1.com gets company1's home page and a call to www.company2.com gets company2's home page, even though they are both the same serve. Anybody know how this is done? Cheers, Gary *************************************************************************** *Gary B. Hoffman, Computing Services Manager e-mail: ghoffman@ucsd.edu* *Graduate School of International Relations and Pacific Studies (IR/PS)* *University of California, San Diego (UCSD) voice: (619) 534-1989* *9500 Gilman Dr., La Jolla, CA 92093-0519 USA fax: (619) 534-3939* *************************************************************************** To: GenWeb@UCSD.EDU From: ghoffman@UCSD.EDU (Gary Hoffman) Organization: IR/PS UC San Diego, La Jolla CA 92093-0519 Date: Sun, 24 Mar 1996 22:50:16 PDT Subject: Virtual Hosting Answer To GenWebbers: I already have my answer to the Virtual Hosting question! Thanks for those replies. Here is a good URL for those who are interested: (Caution, it's pretty geeky.) http://rclsgi.eng.ohio-state.edu/~glaze/virtualhost.html Now, let's see how we can use this for the GenWeb.... Gary *************************************************************************** *Gary B. Hoffman, Computing Services Manager e-mail: ghoffman@ucsd.edu* *Graduate School of International Relations and Pacific Studies (IR/PS)* *University of California, San Diego (UCSD) voice: (619) 534-1989* *9500 Gilman Dr., La Jolla, CA 92093-0519 USA fax: (619) 534-3939* *************************************************************************** From: "W. Wesley Groleau (Wes)" Subject: Re: Stale links between databases. To: genweb@UCSD.EDU Date: Mon, 25 Mar 96 8:37:52 EST In-Reply-To: <1.5.4b12.32.19960323020438.006d38b4@mail.teleport.com>; from "Jeff Murphy" at Mar 22, 96 6:04 pm Mailer: Elm [revision: 70.85] :> From: Jeff Murphy :> Subject: Re: Stale links between databases. :> :> At 12:56 PM 3/22/96 EST, Wes wrote: :> >Tentative example: :> > :> >0 @oiuyt@ INDI :> >1 NAME William Henry /French/ :> >1 NOTE HREF="http://host/cgi-bin/GEDCOM-Match&http://genealogy.org/~wgroleau/WGROLE :> AU.GED&INDI+@oiuyt@> :> >2 CONC Same person in Virginia Isle's database :> :> Hi, Wes. I made a similar suggestion some time ago. First, I believe that :> the only way Well, Jeff, I think your idea is a good one, but I don't think it's similar to mine. If I read you right, you want to keep a file of relations where one side of each relation is unvarying and used as a key for a link, where the other side is variant in presentation, but invariant in what it points to. Seems to me, that this is still subject to the problem others have pointed out of one DB INDI ceasing to exist when the DB owner merges two INDIs. Or one person turning into two due to new info. Plus doesn't this add the risk that the EF may not be as well-maintained as the DB? And if everything else is perfect, there's the inconvenience of a two-step link instead of one. Or is there an HTML function unknown to me that lets a browser make two hops like that with one click? What I was trying to do was have a cgi program look at as much data on the requester's side as possible (actually as much as the programmer coded) and by comparing equivalent fields on the other person's DB, make a judgment as to whether anyone and who is likely to be the same person--then return HTML for that person as if it was a non-stale direct link. Would also be possible (I think) for a cgi script to request (if available) the other person's GEDCOM or a cgi-generated subset (e.g., all INDIs with the same Soundex) and the the requester's program would do the matching. The difference of course, is which person (requester or owner) decides the matching algorithm. Another possible "feature" is to generate a form to select when more than one is a reasonable match. (But here we're in a two-step link again!) One silly mistake in my proposal was the omission of a DB identifier on the serving side. Items needed in the cgi string: 1. URL of matching program 2. URL of requesting DB 3. Info for matching program to find requested INDI in #2 4. URL of DB to be searched for match. What I did NOT propose (i.e., I don't actually endorse my own example) is a "standard" format for all of the above. Just as GEDCOM is "standard" :-) yet genealogical programs that "adhere" to it :-) differ greatly, such a "standard" match-request format would still allow for much flexibility and experimentation in deciding what constitutes a match. I already see nits to be fixed, but I feel they are minor. Perhaps someone else without a beam can remove my mote. -- --------------------------------------------------------------------------- W. Wesley Groleau (Wes) Office: 219-429-4923 Magnavox - Mail Stop 10-40 Home: 219-471-7206 Fort Wayne, IN 46808 elm (Unix): wwgrol@pseserv3.fw.hac.com --------------------------------------------------------------------------- Date: Mon, 25 Mar 96 16:28:22 +0100 From: Anders Andersson Message-Id: <9603251528.AA14933@Mizar.DoCS.UU.SE> To: GenWeb@UCSD.EDU, ghoffman@UCSD.EDU Subject: Re: Virtual Hosting Answer [GenWeb project groups: Project Management, Resource Identifiers] Gary Hoffman writes: >I already have my answer to the Virtual Hosting question! Thanks for those >replies. >Here is a good URL for those who are interested: (Caution, it's pretty >geeky.) No wonder, I'd say, since I consider Virtual Hosting to be a pretty geeky thing (in the `waste of time' category) anyway. > http://rclsgi.eng.ohio-state.edu/~glaze/virtualhost.html >Now, let's see how we can use this for the GenWeb.... Hey, have a look at this neat, recursive, self-modifying assembly language hack! Now, let's see how we can use it for the GenWeb... ;-) While `Virtual Hosting' is a concept which can be applied in a number of contexts, including e-mail where it's standard practice today, I get the impression (and assume for the purpose of this article) that what you primarily have in mind is a WWW URL where GenWeb.Org is found in the server part (like ). If you have other plans, my points below may need some revision. Really, is it necessary? A URL is merely a standard syntax for telling where to find a particular resource, much like the address on a postal envelope is a way to tell where the recipient lives. Implementing a virtual server to provide a set of pages with their own server name is analogous to building a new street to provide a resident with his own street name. Such a move may be defensible for a major company from a marketing point of view, where a neat street address becomes part of the corporate image (like Apple Computer's "1 Infinte Loop" in Cupertino). However, you don't expect to find Rico's Pizza on Rico's Pizza Street, do you? While the domain name space is for all practical purposes unlimited, the numerical IP address space isn't, and the Virtual Hosting method you refer to involves allocating an IP address for an HTTP server, even though the GenWeb Project doesn't own a single piece of hardware! Of course, I'm not afraid that this single address would be the grain tipping the scale, but it would help propagating misconceptions and bad habits in dealing with Internet resources in general. Tell me, which loose organization of people, present or future, would _not_ be eligible for a service-only IP address, if the GenWeb Project is? If you still want to play with this, try and tell me if you think this solves your problem (if there is one). You could even make 130.238.8.14 an A record of WWW.GenWeb.Org! Hackers create all sorts of technical `solutions' without regard to their practical use, and then non-hackers try to apply the solutions to their problems, not always with a successful result. I prefer beginning with the problem and working towards a solution, not the other way around... Considering our limited `budget', I think we should concentrate on the issues which are specific to the GenWeb Project, and avoid spending time on mimicking Internet newbie companies in their PR efforts of `getting on' the net. We were here before most of them. -- Anders Andersson, Dept. of Computer Systems, Uppsala University Paper Mail: Box 325, S-751 05 UPPSALA, Sweden Phone: +46 18 183170 EMail: andersa@DoCS.UU.SE Date: Mon, 25 Mar 1996 07:37:17 -0800 To: genweb@UCSD.EDU From: Jeff Murphy Subject: Re: Stale links between databases. At 08:37 AM 3/25/96 EST, you wrote: >to mine. If I read you right, you want to keep a file of relations where >one side of each relation is unvarying and used as a key for a link, where >the other side is variant in presentation, but invariant in what it points >to. > >Seems to me, that this is still subject to the problem others have pointed >out of one DB INDI ceasing to exist when the DB owner merges two INDIs. All right, let's say a merge occurs. When the data is merged, the notes are merged. Or not. But regardless, won't the links still be valid? Well, they won't if you are linking to a specific record number. That is why there is a need for both a physical and logical record address. The logical address is the one you call invariant. It is assigned to the database, not to the individual. The physical address is connected to the logical address by information not kept in the database. If the information *is* kept in the database, then whoever maintains the database must go through and change every reference when someone else's physical address changes. Makes no sense. That's why there must be some kind of external file. It is the principle upon which data processing is based: file indexes. Logical versus physical pointers. Ah, technology! So, this all works nicely unless the name of the individual is changed. When that happens, the databases are no longer in sync, a match is not found, and users complain loudly, inspiring whoever maintains the database to resolve it. Not all bad. >Plus doesn't this add the risk that the EF may not be as well-maintained >as the DB? I have every confidence that the external file would be far better maintained than the database. The details of maintaining it have yet to be worked out, but I think you will see that as one of the most carefully handled of all these pieces. Besides, the process must of necessity include an update procedure, probably weekly, where everyone receives a fresh copy. And if everything else is perfect, there's the inconvenience >of a two-step link instead of one. Or is there an HTML function unknown >to me that lets a browser make two hops like that with one click? I can't see how you can do a single-step link without including physical addressing in the database. >What I was trying to do was have a cgi program look at as much data on >the requester's side as possible (actually as much as the programmer >coded) and by comparing equivalent fields on the other person's DB, >make a judgment as to whether anyone and who is likely to be the same >person--then return HTML for that person as if it was a non-stale direct >link. Ah. What *I* was trying to do was get the generated html to include the fact that there is another (or more) database(s) which contain the same individual, with the option for the user to jump over to look at the other data. >The difference of course, is which person (requester or owner) decides the >matching algorithm. I guess I can't figure out what you are trying to accomplish. What I want is for people to be able to point to another database from their own. There is nothing that says such links must be reciprocal. Just because I point to your database doesn't mean you will necessarily be pointing to mine. Neither do I think it necessary to show *all* the duplicates. If I find someone who has gone back to Adam on a line that I've gotten to 1850, I'm just going to take my end-of-line person and point to the more complete database. Why include all the information in both databases if the link solves the problem? >Another possible "feature" is to generate a form to select when more >than one is a reasonable match. (But here we're in a two-step link again!) This looks like something totally separate. Are you looking for a way to dynamically decide if there are links, and if so, which are the best? Then you're right, we're not talking about the same thing at all. Jeff Murphy Redmond, Oregon Muhlenberg Co., KY genealogy: http://www.teleport.com/~jmurphy/ subscribe to PAFHELP-L at majordomo@teleport.com Date: Mon, 25 Mar 96 17:31:02 +0100 From: Anders Andersson Message-Id: <9603251631.AA15943@Mizar.DoCS.UU.SE> To: genweb@UCSD.EDU, wwgrol@pseserv3.fw.hac.com Subject: Re: Stale links between databases. [GenWeb project groups: Resource Identifiers, Data Maintenance] Wes writes: >Plus doesn't this add the risk that the EF may not be as well-maintained >as the DB? And if everything else is perfect, there's the inconvenience >of a two-step link instead of one. Or is there an HTML function unknown >to me that lets a browser make two hops like that with one click? You and Jeff may have the picture clear in front of you, but I'm at least partially at loss with what you are trying to do. An example in verbatim might be useful. The EF (`External File') which has been mentioned here would probably serve as a convenient layer between the actual database and the HTML generated to the user. Thus, it doesn't matter how many `steps' the CGI program involved would have to perform in order to produce a URL; the HTML viewer would always see the ultimate URL without even knowing about the EF or the process of resolving the DB link identifier. Thus, if the DB says "wgroleau:oiuyt" and the EF has an entry wgroleau http://genealogy.org/%7Ewgroleau/WGROLEAU.GED&INDI+@~I@ the URL resolver (internal to the DB server) might produce something like William Henry French in the resulting HTML document. Note that the EF doesn't need to be in a format which is globally known; it's essentially up to each server maintainer to decide how to access the contents of the EF and to implement that method. We do need a way to distribute the contents of the EF so that everybody who wants to access the data can get an up-to-date version. One way to do it is via the Domain Name System; another is via Federal Express. Your wording "an HTML function ... that lets a browser make two hops" suggested to me that you may have use for HTTP redirection, which is an almost omnipresent browser feature, but I don't think you need it in this particular case, and I'd like you to pose that question a little more clearly anyway before I try to answer it in detail. -- Anders Andersson, Dept. of Computer Systems, Uppsala University Paper Mail: Box 325, S-751 05 UPPSALA, Sweden Phone: +46 18 183170 EMail: andersa@DoCS.UU.SE From: "W. Wesley Groleau (Wes)" Subject: Re: Stale links between databases. To: genweb@UCSD.EDU Date: Mon, 25 Mar 96 12:17:29 EST In-Reply-To: <9603251337.AA03088@most>; from "W. Wesley Groleau" at Mar 25, 96 8:37 am Mailer: Elm [revision: 70.85] :> :> At 12:56 PM 3/22/96 EST, Wes wrote: :> :> >Tentative example: :> :> > :> :> >0 @oiuyt@ INDI :> :> >1 NAME William Henry /French/ :> :> >1 NOTE :> HREF="http://host/cgi-bin/GEDCOM-Match&http://genealogy.org/~wgroleau/WGROLE :> :> AU.GED&INDI+@oiuyt@> :> :> >2 CONC Same person in Virginia Isle's database And later I corrected: :> One silly mistake in my proposal was the omission of a DB identifier on :> the serving side. Items needed in the cgi string: :> 1. URL of matching program :> 2. URL of requesting DB :> 3. Info for matching program to find requested INDI in #2 :> 4. URL of DB to be searched for match. :> :> What I did NOT propose (i.e., I don't actually endorse my own example) :> is a "standard" format for all of the above. Just as GEDCOM is :> "standard" :-) yet genealogical programs that "adhere" to it :-) differ :> greatly, such a "standard" match-request format would still allow for :> much flexibility and experimentation in deciding what constitutes a :> match. Now I see I need to clarify intent: Suppose the following: (1) I have a database. (2) Someone else has a database (Virginia Isle for example). (3) I happen to know that she and I have a common ancestor, and therefore have SOME overlap in our DB. As many people have already said, if I put a URL in my DB to point to her DB, that URL will probably be incorrect tomorrow. And she may obtain new data making possible a link that I do not have. So instead, I'm suggesting a program that will take whatever facts its programmer thought relevant about my INDI, and about all INDIs in Virginia's DB, and IF a match of high enough probability is found, generate the HTML that would have been generated by the missing or broken URL in the non-CGI examples that were objected to due to their volatility. I fear my hasty typing is not clearing up the obscurity. :-) -- --------------------------------------------------------------------------- W. Wesley Groleau (Wes) Office: 219-429-4923 Magnavox - Mail Stop 10-40 Home: 219-471-7206 Fort Wayne, IN 46808 elm (Unix): wwgrol@pseserv3.fw.hac.com --------------------------------------------------------------------------- Date: Mon, 25 Mar 96 20:04:30 +0100 From: Anders Andersson Message-Id: <9603251904.AA17883@Mizar.DoCS.UU.SE> To: genweb@UCSD.EDU, wwgrol@pseserv3.fw.hac.com Subject: Re: Stale links between databases. [GenWeb project groups: Indexing, Data Maintenance] Wes writes: >Suppose the following: (1) I have a database. (2) Someone else has >a database (Virginia Isle for example). (3) I happen to know that she and >I have a common ancestor, and therefore have SOME overlap in our DB. > >As many people have already said, if I put a URL in my DB to point to her >DB, that URL will probably be incorrect tomorrow. And she may obtain new >data making possible a link that I do not have. I think there are a number of different issues involved here, and that we should (to the extent possible) approach them separately. 1) If you have established that there is some overlap between your database and hers, but not yet determined which individuals match each other (perhaps because, as you suggest, either database may still be subject to extensive revision work), I agree that it makes sense to make a link from your database (as a whole) to hers. You could also make a link from a specific individual in your database to her database, if you want to indicate that this particular person most likely can be found in her database. Both these options require links to databases, as opposed to links to individuals (which I think we have most often implied when talking about `links' in earlier discussion). Then you need a way to specify a query to the target database server, so that you (or the user) can find out which alternatives exist. Since this is a fuzzy match based on randomly chosen data, you cannot take a singular answer for granted, but you must be prepared to handle both a mismatch (no matching person) and a multiple match. Among other things, I think it would be a good idea to inform the user that even a singular link, when produced in this way, is uncertain to some degree. 2) In my opinion, as long as the target database exists and its server responds, your queries will never yield a `broken link'. Instead, a broken link (or broken URL) appears when the database (for whatever reason) goes off-line, the server ceases to exist, or the network connection breaks. This is an entirely different issue, which has to be considered regardless of whether we are talking about queries to resolve fuzzy matches, or established links to well-identified individuals. If both issues require discussion, then I suggest making them separate threads. -- Anders Andersson, Dept. of Computer Systems, Uppsala University Paper Mail: Box 325, S-751 05 UPPSALA, Sweden Phone: +46 18 183170 EMail: andersa@DoCS.UU.SE To: andersa@mizar.docs.uu.se (Anders Andersson), genweb@UCSD.EDU From: ghoffman@UCSD.EDU (Gary Hoffman) Organization: IR/PS UC San Diego, La Jolla CA 92093-0519 Date: Wed, 27 Mar 1996 12:28:30 PDT Subject: Re: Virtual Hosting Answer Anders Andersson,andersa@Mizar.DoCS.UU.SE writes: While `Virtual Hosting' is a concept which can be applied in a number of contexts, including e-mail where it's standard practice today, I get the impression (and assume for the purpose of this article) that what you primarily have in mind is a WWW URL where GenWeb.Org is found in the server part (like ). If you have other plans, my points below may need some revision. Really, is it necessary? ---- Well, Anders, for most people, no. I need to apply virtual hosting here because my employer, the sponsoring organization, wants its home page "at the root". This effectively hides my GenWeb pages because unless you know the path, you can't get there from the root. I just thought if other GenWeb sites established virtual hosts for their major holdings, it would simplify access to the indexes. However, what you write is correct. Any URL will work to reach the index page as long as it's stored in a master index (GenDex) somewhere. Cheers, Gary *************************************************************************** *Gary B. Hoffman, Computing Services Manager e-mail: ghoffman@ucsd.edu* *Graduate School of International Relations and Pacific Studies (IR/PS)* *University of California, San Diego (UCSD) voice: (619) 534-1989* *9500 Gilman Dr., La Jolla, CA 92093-0519 USA fax: (619) 534-3939* *************************************************************************** Date: Thu, 28 Mar 96 19:04:10 +0100 From: Anders Andersson Message-Id: <9603281804.AA14360@Mizar.DoCS.UU.SE> To: genweb@UCSD.EDU, ghoffman@UCSD.EDU Subject: Re: Virtual Hosting Answer [GenWeb project groups: Operations, Resource Identifiers] Gary Hoffman writes: >Anders Andersson,andersa@Mizar.DoCS.UU.SE writes: ... >Really, is it necessary? >Well, Anders, for most people, no. I need to apply virtual hosting here >because my employer, the sponsoring organization, wants its home page "at >the root". This effectively hides my GenWeb pages because unless you know >the path, you can't get there from the root. I just thought if other GenWeb >sites established virtual hosts for their major holdings, it would simplify >access to the indexes. However, what you write is correct. Any URL will >work to reach the index page as long as it's stored in a master index >(GenDex) somewhere. I wouldn't say that your GenWeb pages are "hidden". The canonical way of describing the location of some document or resource on the WWW is via a URL (Uniform/Universal Resource Locator). is no more and no less a URL than is . It could be that one URL is easier to remember than another, but in my opinion URL's are best stored in writing, not in human brain cells, and one URL is just as good as another. I think it's more important that the URL's are stable (so that they are still valid a few months after they were printed in a magazine) than them being mnemonic. Work needs to be done here. There is no reason an arbitrary server should be expected to provide a full index to its entire contents. A service provider may choose to offer server space to a number of mutually unrelated resources, without keeping all of them in a public index. Anyone interested in genealogy will be more likely to encounter the GenWeb project in some of the genalogical indices that already exist elsewhere, than in some list of pages maintained at UCSD. A provider's index is primarily of concern to the provider, who thus can demonstrate affiliation with a lot of happy customers. Virtual hosting for WWW URL's is a PR thing, intended to establish "easy" or "cute" URL's or make an organization appear like it owns and maintains its own computer hardware, making as little reference to their actual Internet provider as possible. You don't need virtual hosting merely because doesn't provide a direct link to the GenWeb project, as there is no reason telling GenWeb users to look at that page in the first place. You should tell people to look at , and they won't even notice whether UCSD provides virtual hosting to anyone. You can use with the same result, since the name DEMO.GENWEB.ORG has been made an alias of FUJI.UCSD.EDU in the Domain Name System, but since the DNS only deals with the server name part and not the entire URL, having an alias doesn't gain you a lot. If you decide to move to a new URL, have you considered how much work it will take to tell everybody your new address? According to Alta Vista, "http://demo.genweb.org/gene" occurs in 181 documents, while "www.genweb.org" of course isn't known anywhere yet. Expect to see the old URL floating around for years, and a lot of people will keep trying to use it. If your problem is that some people today access a URL different from the one you advertise, do you seriously believe that advertising a new URL will solve the problem? Please at least tell us about your plans in detail before implementing them, and we may be able to point out some pitfalls. I admit that is not a perfect URL for the project in the long term, since UCSD may decide that the GenWeb project is a threat to human civilization and throw it out entirely, not even leaving a redirectional link, but that applies to anything on the Internet today. Instead of setting up a virtual host with your own domain name, you can obtain essentially the same effect by employing a generic redirection service, say (the THIS.IS server actually exists, using the Icelandic top-level domain in an unorthodox fashion; see for details). I don't think you have tried yet (suggestions for a domain name for this server are welcome)... -- Anders Andersson, Dept. of Computer Systems, Uppsala University Paper Mail: Box 325, S-751 05 UPPSALA, Sweden Phone: +46 18 183170 EMail: andersa@DoCS.UU.SE Date: Fri, 29 Mar 96 14:39:51 PST Message-Id: <9603292239.AA29362@methane.sri.com> Subject: Re: Sorting indexes by locality To: genweb@UCSD.EDU X-Mailer: AIR Mail 3.X (SPRY, Inc.) As part of my WWW genealogy I generated an index of locations. Also all locations for all events are linked to tables showing all other events that took place at that location. As you will see, any deviation from a strict hierarchical accuracy will cause locations to sort differently and therefore, not link. However, from this perspective these are fairly easy to see and correct. You can view all this at: http://www-cmrc.sri.com/kim/intro.html Hope this helps. Kim Everingham everingham@methane.sri.com <---- Begin Included Message ----> Date: Sun, 24 Mar 1996 13:25:28 -0800 From: Jeff Murphy Subject: Sorting indexes by locality To: genweb@ucsd.edu [Gene, this is mostly to you, I guess, but there are others who it might affect, and I wanted to hit everyone at once.] There was some discussion awhile ago about the possible ability to create an index sorted by location. The need for such a sort has now arisen, and I wanted to discuss it to see if it was possible. I have proposed to the KY researchers that we create a master web page that will take us to the various counties in KY, so that we can look for individuals by location. I am in the process of discussing the options still, but it looks very likely that we can set it all up. Many of us KY researchers have relatives in multiple counties. For instance, while I specialize in Muhlenberg, I have people in the 8 surrounding counties, and some elsewhere. But many people would not search a Muhlenberg database for people from Ohio, Butler, Trigg or other counties. It doesn't occur to them. In setting up a locality index for all names of those in KY (birth, death, burial, marriage), we would enable all those county pages to point to a single index, which would include all the current genweb sites. And yet it would be a subset, as it would not include anything not in KY. We would hope that as new genweb sites and databases are added they would be automatically included in an updated index - much like what Gene is doing now with his comprehensive index. I think it would be best if this were sorted by name within state, and showing the location, so if there are 5 John SMITHs, say, the user could go first to the one from the county nearest their research area. Your responses would be appreciated. Jeff Murphy Redmond, Oregon Muhlenberg Co., KY genealogy: http://www.teleport.com/~jmurphy/ subscribe to PAFHELP-L at majordomo@teleport.com <---- End Included Message ----> From: Scott McGee Received: by wales id ; Fri, 29 Mar 96 19:48:30 CST Date: Fri, 29 Mar 96 19:48:30 CST Message-Id: <9603300148.AA11980@wales> To: elijah-l@genealogy.emcee.com, gedcom-l@vm1.nodak.edu, genweb@UCSD.EDU, lines-l@vm1.nodak.edu, zion@sims.net Subject: bye - for now Friends, I have accepted a job in Salt Lake City, Utah as WebMaster for the Salt Lake Community College. I will be starting there on April 15. I had given my current employer the two weeks notice they asked for (which would have made April 5th my last day), but they got greedy and decided to tell me an hour before quitting time today that they were ending my employment effective today instead. At any rate, with no advance notice, I find myself faced with the immediate loss of most of my computer connectivity. I must therefore unsubscribe from the mail lists I subscibe to. I hope to follow many of them on the few times I can get online through the www archives. I will be back online by the 15th or shortly afterwards. Anyone needing to reach me can email me at smcgee@genealogy.org and I should get it within a day or so most of this time. I will resubscribe when I get my account set up at my new employers. See you again soon Scott Buttered bread always lands butter side * Would YOU mistake these as down (Unless it sticks to the ceiling!) * anyone`s opinions but my own? Email: smcgee@genealogy.org (Scott McGee) Web: http://genealogy.org/~smcgee/homepage.html Date: Fri, 29 Mar 1996 19:24:03 -0800 To: smcgee@genealogy.org From: Jeff Murphy Subject: Congratulations! Cc: genweb@UCSD.EDU Glad you got the job. I assume that they gave you two weeks severance to avoid the obvious lawsuit. (Or you could contact your new employer and ask them if they would be interested in you coming earlier.) I hate it when employers mess around like this. So, let us know when you get back up. I've got kygenweb-l working, and we have had a master page for all KY counties created, and one person now has created the first county page besides mine. Your participation over there might be very helpful. We think of you kindly. :-) You'll have to let me know how to beg to get some of these connectivity ideas implemented. Here I'll have a whole group of Kentuckians - and you know how *they* can be - remember the Alamo? - wanting to tie databases together the way you've done, with a handy search engine and all, and none of the html generators have the facility to show a link to another database. It's a good way to beta test the software on a bunch of pages without worrying too much if it doesn't work the first time. And I suspect that whoever puts it in first is the way we'll go, since nobody is going to want to convert once the data is stored in the notes. We all wish you the very best, and hope you'll have fun with all those Utah Mormons. :-) Once again, congratulations. Jeff Murphy Redmond, Oregon KY genealogy: http://www.teleport.com/~jmurphy/ subscribe to PAFHELP-L or KYGENWEB-L at majordomo@teleport.com Date: Fri, 29 Mar 96 19:30 PST From: smcgee@emcee.com (Scott McGee) To: jmurphy@teleport.com, smcgee@genealogy.org Subject: Re: Congratulations! Cc: genweb@UCSD.EDU Jeff, Actually, no, they offered no severance pay. I guess Iowa law allows that. ( I intend t to find out!) I would be interested in looking at what you are doing, but it might be a rf few weeks first. THanks for the note. Scott GENEALOGY | Do you know who your ancestors are? | Scott McGee -----------+---------------------------------------+--------------------- email: smcgee@genealogy.org | What? Me speak for web: http://genealogy.org/~smcgee/homepage.html | someone else? Nah! ---------------------------------------------------+--------------------- See my genealogy page at http://genealogy.org/~smcgee and my GenWeb page at http://genealogy.org/~smcgee/genweb