Mailing List Archive: 49091 messages
  • Home
  • Script library
  • AltME Archive
  • Mailing list
  • Articles Index
  • Site search
 

dbms3.r

 [1/9] from: g:santilli:tiscalinet:it at: 13-Jan-2002 19:42


Hello Petr! On 13-Gen-02, you wrote: PK> - next Rebol update should also introduce PK> new port type, with ability to perform seek upon direct PK> opened port, maybe you would want to wait for this one? I've been waiting, but now I need it so I'll go with it and improve in the future. PK> The problem with native Rebol dbms is - robustness and PK> stability - locking, several processes working with one db at That's not my goal. I'll use MySQL when I'll need these features; but if I'm writing something like an address book, but complicated enough to require a relational db, I want to use something like dbms.r. PK> library, so really small, available currently for many Really small? Compared to dbms.r? ;-) PK> platforms. Maybe we could adopt that one? It features the PK> same interface as full mySQL version ... Too much complexity. Let's stay simple! PK> PS: of course I don't want you to leave your current thoughts PK> and work, just a suggestion to look at external solution not PK> necessarily reinventing the wheel ... I still think a small REBOL dbms can be useful. I've been using MySQL a lot, and I'm not going to abandon it... but it's not always the best solution IMHO. Regards, Gabriele. -- Gabriele Santilli <[giesse--writeme--com]> - Amigan - REBOL programmer Amiga Group Italia sez. L'Aquila -- http://www.amyresource.it/AGI/

 [2/9] from: petr:krenzelok:trz:cz at: 13-Jan-2002 23:17


Gabriele Santilli wrote:
>Hello Petr! > >I still think a small REBOL dbms can be useful. I've been using >MySQL a lot, and I'm not going to abandon it... but it's not >always the best solution IMHO. >
OK, I can now see your point - however - starting to build dbms system and allowing it to be only one-user-per-time system, ie no locking, no concurency, etc., is a mistake in its roots imo ... btw: what about Elan's dbms system? Wasn't it "distributed"? We should look in its principles then at least ... ... I started to build rebol dbms system in Rebol alpha (pre 1.0 days) and I had it almost working. But then I was not satisfied and wanted even more functionality, so I added another functionality ... and so on ... and so on ... in the end I found out, that what I want to have is what I already had with my XBase systems - so I decided to stop the work - too much complex for Rebol script level, reinventing the wheel.... If e.g. mySQL embedded (libmysqld) is free, small enough, but robust, available in sources and already multiplatform, I can still see the point to have a wrapper, and I will link to it one day, once I proceed with my C lang knowledge. The main advantage is, that it is developed by the whole community of dedicated programmers, so I can concentrate on other things ... Maybe RT could license and pack it into rebol and we would be done :-) Just one question - where can I read more about your rdbms thoughts? I would like to know more about it. For me the one and only view of dbms is that of data consistency. We built our sales department IS in CA-Visual Objects. It used XBase .dbf files and foxpro indices (.CDX). We thought it was robust enough. We put our databases onto Novell Netware server, and our users were connected thru old Token-Ring network The network was problematic. We stopped the testing phase, because our databases became complete mess in less than two weeks. Once you have some even small cross table relationships, and you have not transaction support and roll-backs, you are dead. Inconsistent data = dead data = no data. NEVER AGAIN :-) No data corruption since we switched to Advantage Database Server - the "complexity" outweights "simplicity" in factor more than 1000 times for us .... -pekr-

 [3/9] from: mike:yaunish:home at: 14-Jan-2002 8:16


At 07:42 PM 13/01/02 +0200, you wrote:
>Hello Petr! >On 13-Gen-02, you wrote:
<<quoted lines omitted: 20>>
>MySQL a lot, and I'm not going to abandon it... but it's not >always the best solution IMHO.
I would agree. I have used both dbms.r and MySQL with REBOL and have been fairly disappointed with the slow performance of MySQL. If you need a small database with single user, very fast access then dbms.r is the way to go. As time goes by I am sure we will get a clearer idea where each database system does best. After seeing the performance of MySQL I am keen to see how far (read how big) dbms.r can take me.

 [4/9] from: greggirwin:mindspring at: 14-Jan-2002 13:13


Hi Petr, Just to add another view... << Once you have some even small cross table relationships, and you have not transaction support and roll-backs, you are dead. >> I think that depends on your needs, how you architect a system, and how you use the tools available. I built a medium size system that dealt with complex objects and we used a small ISAM database which allowed us to persist them more easily than using an RDBMS. We had to account for the relationships and protection mechanisms to fit our needs, but it was a much better solution for us, given that we were dealing with a small number of large objects. Many relationships were implicit due to the object hierarchy. We accomplished our goals with semaphores and had very few incidents even though we didn't have explicit transaction support built in to our system. --Gregg

 [5/9] from: g:santilli:tiscalinet:it at: 14-Jan-2002 21:35


Hello Petr! On 14-Gen-02, you wrote: PK> OK, I can now see your point - however - starting to build PK> dbms system and allowing it to be only one-user-per-time PK> system, ie no locking, no concurency, etc., is a mistake in Well, the mistake is in calling it "dbms"... :-) I can change that to THF (Table Handling Functions) or anything seems more appropriate. PK> btw: what about Elan's dbms system? Wasn't PK> it "distributed"? We should look in its principles then at PK> least ... It stores the data in a way similar to dbms.r (a bit better, actually). It has better handling of references between tables, but they need to be hard-coded IIRC. The query system of dbms.r is more powerful IMHO. PK> had with my XBase systems - so I decided to stop the work - PK> too much complex for Rebol script level, reinventing the PK> wheel.... Indeed, but that is not what *I* want --- I already use MySQL when I need it. I'd be satisfied with the current dbms.r if it was able to handle more than 1000 rows per table without starting to be extremely slow (due to the fact that filesystems are not designed to have some thousands of files in a directory... :). PK> ... Maybe RT could license and pack it into rebol and we PK> would be done :-) :-) PK> Just one question - where can I read more about your rdbms PK> thoughts? I would like to know more about it. For me the one [...] Well... it always depends on the goal you have. In your case, even MySQL might be limited (you may need triggers and referential integrity); personally, I think that the relational model is starting to be old. It is very simple, so it is very good; but can we do better? I have some ideas, but nothing that can be put into practice yet. Regards, Gabriele. -- Gabriele Santilli <[giesse--writeme--com]> - Amigan - REBOL programmer Amiga Group Italia sez. L'Aquila -- http://www.amyresource.it/AGI/

 [6/9] from: g:santilli:tiscalinet:it at: 14-Jan-2002 21:21


Hello pat665! On 13-Gen-02, you wrote: p> making a few suggestion ( for example a more secure way to p> delete a record ), helping a bit for the documentation (FAQ, Indeed, my slowness in fixing that bug was one of the reason to begin this. Anyway, if you're still having problems (you shouldn't with the workaound I provided I hope...), let me know and I'll send you a fixed (but not tested) version. Regards, Gabriele. -- Gabriele Santilli <[giesse--writeme--com]> - Amigan - REBOL programmer Amiga Group Italia sez. L'Aquila -- http://www.amyresource.it/AGI/

 [7/9] from: rebol665:ifrance at: 15-Jan-2002 14:30


Hi Gabriele, I have no problem anymore but I'll be glad to test your version. Patrick

 [8/9] from: g:santilli:tiscalinet:it at: 2-Feb-2002 16:40


Hi! FYI, development part 3 has been posted today to the new list. Regards, Gabriele. -- Gabriele Santilli <[giesse--writeme--com]> - Amigan - REBOL programmer Amiga Group Italia sez. L'Aquila -- http://www.amyresource.it/AGI/

 [9/9] from: louisaturk:eudoramail at: 3-Feb-2002 16:03


Sorry to have to ask, but I am not receiving all of the email to this list, nor is all of my email being posted. I am beginning to wonder if I am properly subscribed. How can I find out? Anyway, I am very interest in the new db list. How does one subscribe to it? Louis At 04:40 PM 2/2/2002 +0200, you wrote:

Notes
  • Quoted lines have been omitted from some messages.
    View the message alone to see the lines that have been omitted