From - Sun Oct 05 20:45:50 1997 Return-Path: Received: from punt-1.mail.demon.net by mailstore for rleyton@dogbert.demon.co.uk id 876047914:10:07920:0; Sun, 05 Oct 97 11:38:34 BST Received: from mail.acm.org ([199.222.69.4]) by punt-1.mail.demon.net id aa0916927; 5 Oct 97 11:37 BST Received: from brookes.ac.uk (csmail.brookes.ac.uk [161.73.1.1]) by mail.acm.org (8.8.5/8.7.5) with ESMTP id GAA123254 for ; Sun, 5 Oct 1997 06:42:08 -0400 From: majordomo@brookes.ac.uk Received: by brookes.ac.uk (8.8.7/SMI-SVR4) id LAA22078; Sun, 5 Oct 1997 11:37:44 +0100 (BST) Date: Sun, 5 Oct 1997 11:37:44 +0100 (BST) Message-Id: <199710051037.LAA22078@brookes.ac.uk> To: rleyton@acm.org Subject: Majordomo file: list 'leap' file 'leap.9708' Reply-To: majordomo@brookes.ac.uk X-Mozilla-Status: 2001 -- >From leap-owner Tue Aug 5 00:53:44 1997 Received: by brookes.ac.uk (8.8.4/SMI-SVR4) id AAA24909; Tue, 5 Aug 1997 00:52:51 +0100 (BST) Message-ID: <33E63B23.E92919EE@acm.org> Date: Mon, 04 Aug 1997 21:27:15 +0100 From: Richard Leyton X-Mailer: Mozilla 4.01 [en] (Win95; I) MIME-Version: 1.0 To: leap@brookes.ac.uk Subject: leap list: YABWPRW X-Priority: 3 (Normal) Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-leap@brookes.ac.uk Precedence: bulk Reply-To: leap@brookes.ac.uk (Yet Another Bloody Web Page Re-Write) Yup, I'm at it again... I'd be interested in your comments on the LEAP home page. I've removed frames so it should be easier for people to read, added a copy of the LEAP 1.0 documentation, Updated a few old links. Oh, and I've got a nicer colour scheme... Pastels... :-D Regards, Richard. -- Richard Leyton, Sybase DBA and Unix SA http://www.dogbert.demon.co.uk/home.html mailto:rleyton@acm.org >From leap-owner Tue Aug 12 00:51:21 1997 Received: by brookes.ac.uk (8.8.4/SMI-SVR4) id AAA01112; Tue, 12 Aug 1997 00:49:16 +0100 (BST) Message-ID: <33EF9624.18D854A1@acm.org> Date: Mon, 11 Aug 1997 23:45:56 +0100 From: Richard Leyton X-Mailer: Mozilla 4.01 [en] (Win95; I) MIME-Version: 1.0 To: leap@brookes.ac.uk, rleyton@acm.org Subject: leap list: Forthcoming features in LEAP 1.1.0 X-Priority: 3 (Normal) Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-leap@brookes.ac.uk Precedence: bulk Reply-To: leap@brookes.ac.uk Dear all, A large(ish) note to let you know that I am working, when time permits, on the next release of LEAP. I wanted to give you all an idea of the features that will be forthcoming in version 1.1.0: * Fixed format self containted files, containing: (Partially implemented) o Relation name, temporary status, etc. o Number of attributes o Attribute definition, including data size o Header marker o Tuples. * An optimisation step (Already implemented. Available for 1.0 on request) * Basic remote connections * Command line editor (with history feature) with Emacs and vi key mappings (Already implemented) The thrust behind the first step, of having a binary format file defining an entire relation is to enable indicies to be implemented (Anticipated in LEAP 1.3). It should also dramatically improve performance, as well as improve the code base! There will no longer be seperate .rel, .fld and .tmp files (reduced disk space), LEAP should be far more scalable (LEAP starts having trouble in 103 cardinality relations. deletes/updates will also be possible (although not necessarily in 1.1.0! Depends on time) The optimisation step may not contain any actual optimisation code, but is a minor code reorganisation that will be useful for enabling code to be slotted in if requried. >>This change is available on request to be applied to LEAP 1.0. << Basic remote connections will simply enable as an option LEAP to be started and take input from a socket rather than a keyboard. I'm yet to work out how far to take this approach (RPC's would be a far better solution), but I have some ideas that may make it quite an easy interim solution that allows multiple users running off of one installation. Command line editor using emacs and vi keys is enabled by use of the GNU readline library. This will not be included in the distribution, but is very easily available from gnu sites. If it's not installed, the configure script should pick up on the fact, and will revert to the basic editing available in LEAP 1.0. If you're used to bash, gdb, or any of the gnu utilities, you'll recognise it immediately. If you've used emacs or vi, you'll feel at home. If none of the above, then the cursor keys work just fine as well! I have to say, it makes a _hell_ of a difference to LEAP. What's with the version numbers? I'm taking the same approach for version numbers that was taken in the Linux kernel (Don't worry, no delusions of grandeur...). major.minor.revision.patch * Major - heavy duty change in code base, functionality, etc. * Minor - Odd number releases (1.1, 1.3) are essentially "beta" releases that are not necessarily stable; Even number releases (1.2, 1.4) are stable "full" releases that should be safe to use with no untoward affects. * Revision - In a "beta" release, minor change or improvement to a new development thrust. x.x.0 would be the first stab, and subsequent versions should indicate improvements in stability and functionality. In a proper release it would indicate a fix to an encountered problem * Patch - Few and far between, I hope, but emergency fixes mainly. Anyway, that's enough for now. The above will be on theBRAND NEW web site in due course, but you saw it hear first... :-) Some of the items above are very specific improvements, but I am always happy to hear your comments on the direction you think LEAP should be going. I'd also very much like to hear your thoughts on what I've layed out above, either directly, or via the list to perhaps provoke a bit of discussion on LEAP's future direction. Regards, Richard. -- Richard Leyton http://www.dogbert.demon.co.uk/leap.html mailto:rleyton@acm.org My opinions, My mistakes. >From leap-owner Fri Aug 15 00:07:47 1997 Received: by brookes.ac.uk (8.8.4/SMI-SVR4) id AAA07315; Fri, 15 Aug 1997 00:05:01 +0100 (BST) Message-ID: <33F35553.3C47C8E6@acm.org> Date: Thu, 14 Aug 1997 19:58:28 +0100 From: Richard Leyton X-Mailer: Mozilla 4.01 [en] (Win95; I) MIME-Version: 1.0 To: leap@brookes.ac.uk Subject: leap list: Bug fix for LEAP 1.0 X-Priority: 3 (Normal) Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-leap@brookes.ac.uk Precedence: bulk Reply-To: leap@brookes.ac.uk Dear all, It's been brought to my attention that you may suffer from segmentation faults due to a problem in the relation.c module. Sven Verdoolaege has kindly provided me with the following patch which I have checked out and found to be ok. To apply it, save this e-mail to [filename], and copy it to leap-1.0/src then type: patch < [filename] The technical explanation is that LEAP 1.0 tries to close a file (indicating that the relation is temporary) that has never been opened. This can therefore lead to segmentation faults. It does not appear to be fatal in all libc's, but the GNU library glibc 2.0.4 is known to cause this. If you have such problems in a version of LEAP 1.0, apply this patch. This patch is obsoleted by the work I am doing on LEAP 1.1, due to a completely new relation format. The temporary status is a flag in the header of the relation file. For more details, see the LEAP home page. This patch will be available for download off of the web page in the near future. Regards, Richard. *** relation.c.orig Thu Aug 14 19:37:47 1997 --- relation.c Thu Aug 14 19:32:11 1997 *************** *** 523,533 **** rel->temporary=FALSE; } else { rel->temporary=TRUE; } - /* Close the file */ - fclose(tmp); - relation_full_path(rel,temp); strcat(temp,LEAP_HASH_EXT); tmp=fopen(temp,"r"); --- 523,532 ---- rel->temporary=FALSE; } else { rel->temporary=TRUE; + /* Close the file */ + fclose(tmp); } relation_full_path(rel,temp); strcat(temp,LEAP_HASH_EXT); tmp=fopen(temp,"r"); *************** *** 550,555 **** --- 549,557 ---- if (status_trace) { printf(" Done.\n"); } + + /* Close the temporary file */ + fclose(tmp); } else { /* No table exists, so create it... */ *************** *** 567,575 **** printf(" Done.\n"); } - - /* Close the temporary file */ - fclose(tmp); /* Reset the Hash Table TODO - DONE 06.05.1996 */ /* rel->hash_table=NULL; */ --- 569,574 ---- >From leap-owner Fri Aug 15 00:15:51 1997 Received: by brookes.ac.uk (8.8.4/SMI-SVR4) id AAA07886; Fri, 15 Aug 1997 00:13:56 +0100 (BST) Message-ID: <33F38C2D.34F07863@acm.org> Date: Thu, 14 Aug 1997 23:52:29 +0100 From: Richard Leyton X-Mailer: Mozilla 4.01 [en] (Win95; I) MIME-Version: 1.0 To: LEAP Mailing List Subject: leap list: News on LEAP 1.1.0 X-Priority: 3 (Normal) Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-leap@brookes.ac.uk Precedence: bulk Reply-To: leap@brookes.ac.uk Well, to be honest I didn't expect to be sending out a mail like this for some time, but then I've made some substantial progress in the last day or two with irratating minor bugs that looked more serious than they were. Anyway, all of the relational operators in 1.1.0 are using the new relation format and the news I wanted to let you know was that the peformance on large relations is _significantly_ better. In LEAP 1.0 and earlier, if you tried to put a large relation (with more than a few hundred tuples in) through LEAP, LEAP would use excessive amounts of memory, crash, take a long time, or all of them together. Whilst hash files are not configured in the same fashion in the newer version of LEAP, the performance is significant. I created a relation with two attributes, one a fixed string of 255 chars containing the world "string" in all tuples, the other a sequential integer number, different in each tuple. I then generated exactly 255624 entries. I then executed: project (count) (B) to project out the "string", which should return just one tuple. It did this correctly as expected. What I didn't expect was the time it would take. It took between 17.4 seconds and 19.97 seconds in the trials I tried. Rough calculation gives a throughput of between 14648.94 tuples/sec and 12800.40 tuples/sec. Ok, I know, this is early days, and we're hardly talking large degree relations here. But I'm very exited by these early results. The new format relations are both faster, more scalable and more flexible, and the function calls have not changed too much at all. Some other stats.... select (count) (A<"100") read: 255624 tuples wrote: 99 tuples time: 23.03 seconds throughput: 11099.61 tuples/second My hardware, in case you're wondering is a Linux 2.0 machine with a Pentium Pro 200Mhz CPU and 32Mb RAM. The relation in question is 68Mb in size, so at best approximately half of the file could be in the file system cache. I flushed the cache to try and get a worse case, and that's the 19.97s figure... Anyway, this news is good news, and I'm more optimistic now than ever that I'll be able to release a version soon (maybe before the end of August). If you're interested in receiving an early copy, drop me a line. I know there are going to be a few bugs that need ironing out, and I'd appreciate any help I can get. Regards, Richard. -- Richard Leyton http://www.dogbert.demon.co.uk/home.html mailto:rleyton@acm.org My opinions, My mistakes. >From leap-owner Fri Aug 15 00:33:45 1997 Received: by brookes.ac.uk (8.8.4/SMI-SVR4) id AAA08180; Fri, 15 Aug 1997 00:31:48 +0100 (BST) Message-ID: <33F3913E.4D1CF619@acm.org> Date: Fri, 15 Aug 1997 00:14:06 +0100 From: Richard Leyton X-Mailer: Mozilla 4.01 [en] (Win95; I) MIME-Version: 1.0 To: LEAP Mailing List Subject: leap list: The only one? X-Priority: 3 (Normal) Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-leap@brookes.ac.uk Precedence: bulk Reply-To: leap@brookes.ac.uk LEAP, if you were not aware, has an entry on Yahoo. More recently it has been classified as the only UK RDBMS: http://www.yahoo.co.uk/Computers_and_Internet/Software/Databases/ On a more serious note, the LEAP web page has a page on "Related Sites". If there is a site you think would be useful for the casual browser, please let me know. People might well click on LEAP expecting a serious RDBMS for free... If there is something you can recommend (along the lines of msql, for example), please let me know and I'll add it. Regards, Richard. -- Richard Leyton http://www.dogbert.demon.co.uk/home.html mailto:rleyton@acm.org My opinions, My mistakes. >From leap-owner Tue Aug 19 21:09:26 1997 Received: by brookes.ac.uk (8.8.4/SMI-SVR4) id VAA25142; Tue, 19 Aug 1997 21:04:09 +0100 (BST) Message-ID: <33F9F4D2.9CC60C62@acm.org> Date: Tue, 19 Aug 1997 20:32:34 +0100 From: Richard Leyton X-Mailer: Mozilla 4.01 [en] (Win95; I) MIME-Version: 1.0 To: LEAP Mailing List Subject: leap list: [Fwd: [ANNOUNCE] Release of LEAP 1.0 32bit binary] X-Priority: 3 (Normal) Content-Type: multipart/mixed; boundary="------------70A7FCF1BBF4442F1E93374C" Sender: owner-leap@brookes.ac.uk Precedence: bulk Reply-To: leap@brookes.ac.uk This is a multi-part message in MIME format. --------------70A7FCF1BBF4442F1E93374C Content-Type: text/plain; charset=us-ascii Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Content-Transfer-Encoding: 7bit For your information, this mail has just gone to: comp.edu comp.databases.theory comp.databases -- Richard Leyton http://www.dogbert.demon.co.uk/home.html mailto:rleyton@acm.org My opinions, My mistakes. --------------70A7FCF1BBF4442F1E93374C Content-Type: message/rfc822 Content-Transfer-Encoding: 7bit Content-Disposition: inline Path: news.demon.co.uk!dispatch.news.demon.net!demon!dogbert.demon.co.uk!not-for-mail From: Richard Leyton Newsgroups: comp.databases.theory,comp.databases,comp.edu Subject: [ANNOUNCE] Release of LEAP 1.0 32bit binary Date: Tue, 19 Aug 1997 20:18:33 +0100 Message-ID: <33F9F189.FA5BBA47@acm.org> NNTP-Posting-Host: dogbert.demon.co.uk X-NNTP-Posting-Host: dogbert.demon.co.uk [194.222.192.194] X-Mailer: Mozilla 4.01 [en] (Win95; I) MIME-Version: 1.0 X-Priority: 3 (Normal) Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Xref: demon comp.databases.theory:7068 comp.databases:27953 comp.edu:9657 This is to inform you that a 32bit console binary version of the LEAP RDBMS system has been made, and may be downloaded from: ftp://ftp.demon.co.uk/pub/compsci/databases/leap/wleap32bit1_0.tar.gz Previously LEAP has been only available for Unix platforms, and in a 16bit Windows version. Full 'C' source code is included under the GNU General Public License. Full information, as well as a link to the above ftp site, is available on the LEAP web site: http://www.dogbert.demon.co.uk/leap.html http://www.brookes.ac.uk/~e0190404/leap.html LEAP is a free RDBMS (Relational Database Management System) that has been used as an educational tool in Universities around the world to aid the students as they learn database theory. It has also been used as a light weight back end database system. LEAP is in active development by the author. LEAP 1.1.0 will be available for beta testing within the next month, and _significantly_ improves the capacity and performance of LEAP, as well as adding some significant new functionality. This will enable LEAP to make significant steps toward becoming not only an educational tool, but also a reliable back end RDBMS. Entirely for free. To stay informed on the new LEAP developments, a mailing list is available. To subscribe, send a one line e-mail to "majordomo@brookes.ac.uk" containing, in the body of the message, "subscribe leap". Regards, Richard Leyton. -- Richard Leyton http://www.dogbert.demon.co.uk/leap.html mailto:rleyton@acm.org My opinions, My mistakes. --------------70A7FCF1BBF4442F1E93374C-- >From leap-owner Tue Aug 26 00:00:37 1997 Received: by brookes.ac.uk (8.8.4/SMI-SVR4) id XAA00787; Mon, 25 Aug 1997 23:59:59 +0100 (BST) Message-ID: <3401A676.E3EC2769@acm.org> Date: Mon, 25 Aug 1997 16:36:22 +0100 From: Richard Leyton X-Mailer: Mozilla 4.01 [en] (Win95; I) MIME-Version: 1.0 To: LEAP Mailing List Subject: leap list: OS/2 testers X-Priority: 3 (Normal) Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-leap@brookes.ac.uk Precedence: bulk Reply-To: leap@brookes.ac.uk Hi all... Just a quicky... Is anybody out there prepared to test an OS/2 version of LEAP 1.0? Just had an e-mail from a group of people offering to do the port, and as it's outside of my control, I'd like to have somebody with OS/2 experience/access who could check it out before I release it... Richard. -- Richard Leyton http://www.dogbert.demon.co.uk/home.html mailto:rleyton@acm.org My opinions, My mistakes. >From leap-owner Fri Aug 29 00:14:59 1997 Received: by brookes.ac.uk (8.8.4/SMI-SVR4) id AAA22914; Fri, 29 Aug 1997 00:14:32 +0100 (BST) Message-ID: <340602C6.977980EF@acm.org> Date: Thu, 28 Aug 1997 23:59:18 +0100 From: Richard Leyton X-Mailer: Mozilla 4.01 [en] (Win95; I) MIME-Version: 1.0 Followup-To: comp.os.linux.development.apps To: comp.os.linux.development.apps@acm.org, comp.databases@acm.org CC: LEAP Mailing List Subject: leap list: How to make a DBMS more useful to developers... X-Priority: 3 (Normal) Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-leap@brookes.ac.uk Precedence: bulk Reply-To: leap@brookes.ac.uk Dear all, Over the last two to three years, I've been developing a relational database management system that has, until now, been geared explicitly at students as a teaching aid for their study of database theory. The RDBMS is used by an ever increasing number of universities and colleges around the world, and has been available for the last 18 months, under the GNU General Public license. As far as it goes, it's been a great success. Recently, however, I've received a number of requests, primarily from members of the Linux community, asking if my system (LEAP) could be used as a back end dbms for various applications and systems. Until now, I've always replied that it wasn't really geared up to such a task. A linkable library (in any shape or form) didn't exist, it's performance was abysmal to say the least, and such a use was not really a priority of mine (until now). Over the last month or two, I've been throwing ideas around about how to progress the system, and I have therefore been looking at ways of enabling the dbms to be _easily_ used as a back-end dbms in your systems. The purpose of this posting is therefore to ask you, developers (and users?) of applications in Linux, what you would like to see from a back end dbms that you could easily encorporate into your applications. Over the last month or so, I've been working towards something along the lines of: * Clearly defined API. * Shared and static library compilation options. * An accessable plain text query language that removes the need to navigate the database structure. ie. if you're implementing a system, you don't care _how_ the data is stored (ie. whether indexes are used or n ot), just the fact that a query of the form "select (customers) (id='1234')" returns a specific customer record in a reasonable time. The dbm library, for example, does not allow such a freeform notation (to my knowledge). * Half-way decent performance on ufs file systems with no internal caching of data (as yet). * Internal data types that correspond well to C/C++ data types. Specifically, I'm after answers to questions such as: * Do you _really_ want SQL support? This is not something LEAP currently supports for historical, scope and performance reasons. I'm also (currently) not convinced it something that is significantly important to an application developer wanting a simple means to store data for subsequent retrieval in a non-portable (between back-end systems, such as rdbms servers) context. * Is multi-connection support a significant must-have, ie. are multiple instances of your application (whatever it might be) likely to want to access the same database storage structure(s). (Seperate processes or remote servers). * Would you want low-level access to query operators (such as select, update, delete), or be happy with higher level operators, such as process the following query "delete (table) (id=1234)" and return the error code, or a mixture of both? * What you really want from a dbms behind your application. If you're interested in what I've mentioned and want to find out more, please take a look at the home page of LEAP to get an idea of the system _to date_. Links are available to the many ftp mirror sites: http://www.dogbert.demon.co.uk/leap.html http://www.brookes.ac.uk/~e0190404/leap.html _Please_ bear in mind that the home page reflects the version of the software that is available now (1.0). The version currently under development is a rapidly mutating beasty. Beta releases of version 1.1 will be available soon, but based on replies to my ramblings above, it will be changing even more significantly (to be in line with popular requirements). My medium term goal is to make the 1.3 release the version that most appeals to developers for incorporate into their applications, and move on from there. The whole purpose of this posting is to therefore to try and make sure that what is worked on over the next few months is something of use to the Linux/GNU community of application developers. What do you want from a library that would handle your data storage requriements. Please feel free to drop me _any_ suggestions, comments, criticisms, flames, and so forth you might have, either via e-mail, or as a reply to this posting on comp.os.linux.development.apps. Looking very much forward to hearing from you soon, Richard Leyton. PS. If you're interested in staying in touch with what is going on with LEAP, there is a mailing list. Send a one line e-mail to: majordomo@brookes.ac.uk with the line "subscribe leap" in the _body_ of the e-mail. Full details are, of course, on the above home page(s). -- Richard Leyton http://www.dogbert.demon.co.uk/leap.html mailto:rleyton@acm.org My opinions, My mistakes. >From leap-owner Fri Aug 29 06:30:55 1997 Received: by brookes.ac.uk (8.8.4/SMI-SVR4) id GAA02349; Fri, 29 Aug 1997 06:30:41 +0100 (BST) Date: Wed, 27 Aug 1997 21:28:24 GMT From: Paul@tmsl.demon.co.uk (Paul Thomas) Message-Id: <462@tmsl.demon.co.uk> To: leap@brookes.ac.uk Subject: Re: leap list: OS/2 testers X-Mailer: PCElm 1.11 Lines: 18 Sender: owner-leap@brookes.ac.uk Precedence: bulk Reply-To: leap@brookes.ac.uk In message <3401A676.E3EC2769@acm.org> Richard Leyton writes: > Hi all... > > Just a quicky... Is anybody out there prepared to test an OS/2 version > of LEAP 1.0? Just had an e-mail from a group of people offering to do > the port, and as it's outside of my control, I'd like to have somebody > with OS/2 experience/access who could check it out before I release > it... I did an OS/2 port of ver 0.11 a while back - it was very easy. In fact the few changes I made were for the Watcom compiler not the os. Anyway, I'm an experienced OS/2 professional programmer and I'd be happy to test the OS/2 version of 1.0 for you. BTW, any plans to add an API interface to Leap? -- Paul Thomas >From leap-owner Fri Aug 29 13:42:22 1997 Received: by brookes.ac.uk (8.8.4/SMI-SVR4) id NAA21023; Fri, 29 Aug 1997 13:39:40 +0100 (BST) From: defaultuser@domain.com Message-ID: <3406D149.5ED1AB33@domain.com> Date: Fri, 29 Aug 1997 14:40:25 +0100 X-Mailer: Mozilla 4.01 [en] (WinNT; I) MIME-Version: 1.0 To: leap@brookes.ac.uk Subject: leap list: Leap X-Priority: 3 (Normal) Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-leap@brookes.ac.uk Precedence: bulk Reply-To: leap@brookes.ac.uk Hello list i have a question about the last Version of Leap for Windows 95/NT why can i not create a database using the create command like it worked on the 0.10 beta version. How can i integrat my old databases in the new version????? thanks for reading Michael >From leap-owner Fri Aug 29 14:39:16 1997 Received: by brookes.ac.uk (8.8.4/SMI-SVR4) id OAA25057; Fri, 29 Aug 1997 14:37:17 +0100 (BST) Message-Id: From: "Leyton, Richard" To: "'leap@brookes.ac.uk'" Subject: RE: leap list: Leap Date: Fri, 29 Aug 1997 14:33:17 +0100 X-Priority: 3 X-Mailer: Internet Mail Service (5.0.1458.49) Sender: owner-leap@brookes.ac.uk Precedence: bulk Reply-To: leap@brookes.ac.uk Mmm. I suspect Michael may not receive this judging by the reply-to address. Nonetheless: The create database functionality was something I never quite got around to before 1.0, because of the wonderful ndir/dirent portability problems. However, I will endevour to get it into 1.1.0, now that I'm a bit of a ndir-meister, so to speak. In the meantime, the way to create a new database is to copy the model database (and all its subdirectories) in NT/Windows explorer to a new directory at the same level (You must have a relship relation in each db). Then, in the master database, to add the database to the database definition relation (leapdb or somesuch... can't remember its exact name right now - it's the other system table other than relship) with "add (leapdb) (dbname, dbname)" To bring over a database from 0.10 (The DOS version. No Windows/NT version exists at version 0.10), you should simply copy all the .rel and .fld files across into the dbname/relation directory. I'll try and double check over the weekend myself. Richard. -- Richard Leyton, Data Services SA/DBA x6503 Salomon Brothers International Ltd, +44 (0)171 721 6503 mailto:richard.leyton@sbil.co.uk > ---------- > From: defaultuser@domain.com[SMTP:defaultuser@domain.com] > Sent: August 29 1997 14:40 > To: leap@brookes.ac.uk > Subject: leap list: Leap > > > > Hello list > > > i have a question about the last Version of Leap for Windows 95/NT > > why can i not create a database using the create command like it > worked > on the 0.10 beta version. > How can i integrat my old databases in the new version????? > > thanks for reading > Michael > >From leap-owner Fri Aug 29 15:08:05 1997 Received: by brookes.ac.uk (8.8.4/SMI-SVR4) id PAA27045; Fri, 29 Aug 1997 15:07:51 +0100 (BST) From: fc@ca.sandia.gov Message-Id: <199708291408.HAA20339@ca.sandia.gov> Subject: Re: leap list: How to make a DBMS more useful to developers... To: leap@brookes.ac.uk Date: Fri, 29 Aug 1997 07:08:40 -0700 (PDT) In-Reply-To: <340602C6.977980EF@acm.org> from "Richard Leyton" at Aug 28, 97 11:59:18 pm X-Mailer: ELM [version 2.4 PL25] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-leap@brookes.ac.uk Precedence: bulk Reply-To: leap@brookes.ac.uk In reference to the following view expressed by Richard Leyton: ... > Over the last month or two, I've been throwing ideas around about how to > progress the system, and I have therefore been looking at ways of > enabling the dbms to be _easily_ used as a back-end dbms in your > systems. > > The purpose of this posting is therefore to ask you, developers (and > users?) of applications in Linux, what you would like to see from a back > end dbms that you could easily encorporate into your applications. Over > the last month or so, I've been working towards something along the > lines of: > > * Clearly defined API. > * Shared and static library compilation options. > * An accessable plain text query language that removes the need to > navigate the database structure. ie. if you're implementing a > system, you don't care _how_ the data is stored (ie. whether > indexes are used or n ot), just the fact that a query of the form > "select (customers) (id='1234')" returns a specific customer record > in a reasonable time. The dbm library, for example, does not allow > such a freeform notation (to my knowledge). > * Half-way decent performance on ufs file systems with no internal > caching of data (as yet). > * Internal data types that correspond well to C/C++ data types. > > Specifically, I'm after answers to questions such as: > > * Do you _really_ want SQL support? This is not something LEAP > currently supports for historical, scope and performance reasons. > I'm also (currently) not convinced it something that is > significantly important to an application developer wanting a > simple means to store data for subsequent retrieval in a > non-portable (between back-end systems, such as rdbms servers) > context. SQL support would make the software very easy to interface to almost everything else available to access databases - in my mind this makes it key to the success of your efforts. I for one would be willing to pay a substantial amount of money for a decent Linux-based database with SQL access. > * Is multi-connection support a significant must-have, ie. are > multiple instances of your application (whatever it might be) > likely to want to access the same database storage structure(s). > (Seperate processes or remote servers). It is fairly easy to implement with an empty interface program that sits on a port and sequentializes the requests. Perhaps if you simply provided this with some sort of access control feature it would suffice - but then long queries would end up lengthening the queue. > * Would you want low-level access to query operators (such as select, > update, delete), or be happy with higher level operators, such as > process the following query "delete (table) (id=1234)" and return > the error code, or a mixture of both? More function is better. > * What you really want from a dbms behind your application. Speed, ease of interface, full-functionality. I would also like to see a variation allowing object-oriented capabilities that also operate from the relational database view. Thus allowing both functionalities. Not likely - eh? FC --- Fred Cohen can be reached at tel:510-294-2087 fax:510-294-1225 >From leap-owner Fri Aug 29 22:26:49 1997 Received: by brookes.ac.uk (8.8.4/SMI-SVR4) id WAA16503; Fri, 29 Aug 1997 22:25:19 +0100 (BST) Message-ID: <340695BA.DD389D7@juno.com> Date: Fri, 29 Aug 1997 05:26:18 -0400 From: "William R Royall Jr." X-Mailer: Mozilla 4.01 [en] (Win95; U) MIME-Version: 1.0 To: leap@brookes.ac.uk Subject: leap list: Re: I haven't had Algebra for 3 years! X-Priority: 3 (Normal) Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-leap@brookes.ac.uk Precedence: bulk Reply-To: leap@brookes.ac.uk I haven't had a corse in Algebra in 3 years. I am in a Calculus class now and we are reviewing Algebra since we will need it. I need to know how to graph parabolic equations ect. I don't remember. Here are a few examples. If you know how to do these please explain how you got the answers :-). f(x)=x^2 - 3 f(x)=x^3 + 2 I would sincerly appreciate any help by Monday night!!!!!! Thanks alot!!!!! Sincerly, William R Royall Jr. >From leap-owner Sat Aug 30 02:02:43 1997 Received: by brookes.ac.uk (8.8.4/SMI-SVR4) id CAA22097; Sat, 30 Aug 1997 02:02:11 +0100 (BST) Message-ID: <34067E4F.72F741BE@acm.org> Date: Fri, 29 Aug 1997 08:46:23 +0100 From: Richard Leyton X-Mailer: Mozilla 4.01 [en] (Win95; I) MIME-Version: 1.0 To: LEAP Mailing List Subject: leap list: LEAP 1.1.0 - More features... X-Priority: 3 (Normal) Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-leap@brookes.ac.uk Precedence: bulk Reply-To: leap@brookes.ac.uk Dear all, Well, I've been busy again with LEAP, and have a few more features coming soon in LEAP 1.1.0. Here's the full list of goodies so far. Feel free to drop any further suggestions... *Fixed format self containted files, containing: (Partially implemented) * Relation name, temporary status, etc. * Number of attributes * Attribute definition, including data size * Header marker * Tuples. *An optimisation step (Already implemented) *Basic remote connections *Tuple deletion (Already implemented) *Command line editor with Emacs and VI keys (Already implemented) *Session recording (Already implemented). *Documented and easy to use API (Planned over two minor releases) The new inclusions (since my previous announcement) are: * Tuple Delete - The format of the delete command is similair to the select operator, ie. "delete (relation) (id='value')" - This has been made possible by the progress I've made of fixed format binary relation files. * Session Recording - Something many of you have asked for over the life span of LEAP since moving from DOS to 'C' and Unix. Well, it's here at last... * API efforts - I sent the request for feedback to the list... This will reflect what I get from that. My goal is simply to make LEAP easy to encorporate as a back end dbms for applications. Release date? Not August... Probably mid September. There are still a fair few bugs in the fixed format relation files I need to iron out, and I'm trying to decidet whether the conversion process should be incorporated into the binary (ie. it detects old format relations, and converts them), or whether a seperate prog should do that as a once off. I'm leaning toward the latter at the moment. Regards, Richard. -- Richard Leyton http://www.dogbert.demon.co.uk/home.html mailto:rleyton@acm.org My opinions, My mistakes.