Re: [Code] Native Database Driver Support

From: Michael Lemler (
Date: 04/02/99

> i dont know what kind of enterprise wide application you are using but
> odbc isnt anywhere near efficient enough for mine.  its okay for one
> time deals where you can use some other app(ie ms access) for your

Depends on the application and the approach.. as with everything in
computers...  Also depends on your backend, network load, network design,
etc.. I my self have had no problem with ODBC and know lots of various
applications that use ODBC as their connectivity agent.  In the world of
computers, there is no one right answer.. you always give and take.  I my
self don't use MS products for ODBC, so I don't know how it performs
other than what I have heard 2cnd hand.  I find that other implementations
of stuff MS uses to be faster (LanManager -- Samba).

> i'm not disagreeing with you that odbc is probably okay for mud
> development. but lets explore some more options. personally i wont play
> a mud with server lag.

The server probably loads the information upon boot... I have not checked
the code myself to see how the circle implementation is.

> i think it would be fairly simple (not to be confused with easy) to
> write a [mud api] interface for the datbase stuff. as many people as

Yeah, it would be simple, but time consuming.. depending on the language
used.  It would be sort of like perl's DBI interface.

Some issues:

Do you want to only support a handful of databases or do you want to
attempt to conform to a standard?  Once the developer decides that, it
will be a simple task from there on out.  I, my self, don't care.  I will
make it do what I want it to do for whatever purposes.

Another issue, what happens when the database drivers are upgraded and
they don't have 100% backwards compatability?  The design of the mud would
have to compensate for that and include the new libs at compile time,
which might effect some other morsle of code.

Also, in an enterprise, you most likely have control over the application
and it's platforms.  With circle, people run in on anything from
UltraSparcIIs with solaris to a Windows 95 box.  Another reason for ODBC.

> autoconf stuff.(unless you had somebody using two rdbms's on their box:
> ie postgre and mysql)

Having autoconf probe for whatever database one is using would not be a
good approach.  I'd rather see a compile time option or something in a
config.c.  Especially if you do compilations on a development box.

Michael Lemler
Systems Engineer and Unix Systems Administrator

     | Ensure that you have read the CircleMUD Mailing List FAQ:  |
     |  |

This archive was generated by hypermail 2b30 : 12/15/00 PST