At 11:53 am -0600 12/5/01, Scott Raney wrote:

>But what we'd *really* like to see is for other ISVs to license the
>engine to build products for markets that MetaCard only tangentially
>competes in now.  The one that's currently really crying out for
>something like this is K12 education, where the dominant multimedia
>authoring products like HyperStudio and Digital Chisel are suffering
>from severe neglect by their owners and are rapidly losing market
>share to psuedo-alternatives like Powerpoint and HTML editors.
>Another example is end-user database products like Filemaker and
>Access which because they lack a decent scripting language are much
>more limited than something like MetaCard + a database back end would
>be.  Other opportunities include CBT (Computer Based Training) and
>HTTP server add-ons (though Simon himself at least is working hard on
>the latter ;-)

Both for K12 and CBT products, database support is probably a big 
requirement (for backend administration anyway). I thought it might 
be useful to list up what the current database options are. Like 
others I guess, my knowledge is pretty spotty in this area, and I'd 
like to know I'm not overlooking any possibilities. It also might 
help any "taker upper" of Scott's suggestion to see what's needed.

To kick off:

1.  Metacard native options.
Metacard stacks provide an easy way to structure and retrieve data. 
The main problem is the memory limitation as stacks are read entirely 
into memory. Not really an option for large databases.

Use a combination of Metacard stacks and external files. Metacard can 
read/write external files, including binary data. Has anyone explored 
the limits of building a database system entirely out of Metacard and 
external files? I've done it with a file of 20,000 records containing 
about 6 fixed-length fields, and building a crude index in a Metacard 
stack. The results were fine for my needs, but I wonder if anyone has 
gone further, e.g. variable length fields, multiple tables, etc.  I'm 
guessing it's not a viable option, but have no foundation for that 
opinion. Best to leave no stone unturned, right?

2.  Valentina
Has externals that work with Metacard on Windows and Mac. Linux 
version reported coming later this year. My (very) limited testing 
showed it to be a good performer on Windows. I don't think it's 
multiuser at present (I may be wrong), but I guess using a Metacard 
app as a server, it's possible to achieve this with sockets. It has a 
pricing structure that seems suitable for most Metacard developers.

3.  FileMaker
It's possible to connect with FileMaker using its "Web Companion" 
feature. FileMaker acts as a "cgi server". It can return data in xml 
format (as well as html). I played with this at the weekend, and was 
surprised how easy it was to set up. Using Metacard as an http 
client, you can perform most FileMaker data functions in this way. 
Currently Mac and Windows only, but a Linux version (server only??) 
was just announced for release in the summer. Licensing may be an 
issue. Connections are limited on the basic version. The "unlimited 
connections" web version costs about $1000.

4.  MySQL, Postgresql,  and other open source databases
I believe you can connect with these using php, perl scripts, etc. 
and by using Metacard as an http client. How effective is this? Would 
it be better if Metacard had its own direct interface to these 
databases?

Are there problematic licensing issues if you wanted to include these 
databases as part of a project? (GPL etc.)

5.  The scarier stuff (Oracle, DB2, MS SQL Server, and other products 
marketed by men with nice suits and bulging eyes)
How scary are these things? A little voice somewhere seems to be 
saying, "stay away". They conjure up images of racks of server, 
armies of consultants, and licensing fees that would bankrupt most 
countries. But for some projects (e.g. CBT in a corporate environment 
where everything has to go on the "house" database) it's probably 
necessary to deal with these things.

What does it need to interface with from Metacard? With a single 
external, is it possible to deal with all ODBC compliant databases? 
Are there other less direct approaches? (as with MySQL, for example).

All additions, corrections, experiences, and related comments welcome

Cheers
Dave Cragg






Archives: http://www.mail-archive.com/[email protected]/
Info: http://www.xworlds.com/metacard/mailinglist.htm
Please send bug reports to <[EMAIL PROTECTED]>, not this list.

Reply via email to