[Biodevelopers] what databases are you using?

Thu Jul 11 09:32:46 EDT 2002

I am not sure about perl's XML serialization, but the package that I use
allows for the mapping of Java objects to XML to be arbitrary.  Though it
does take more work on the developer's behalf if a mapping other than the
default is used.  If you are interested, check out EXML by The Mind
Electric.

I agree that the mapping must be language-neutral.  For instance, NCBI's
blast XML format does not map perfectly to Java objects using a default
mapping.  However, there is no doubt in my mind that any XML schema can be
mapped to a Java (or perl, or any other language) object.  Working with XML
in its raw elements is difficult and often unreasonable for many
applications.  Thus, I think this mapping is essential to application
development.  Programmers do not think in XML, they think in objects,
structs, hashes, or whatever.

As to non-optimal, what do you mean?  When I map objects to XML, it is not
as just a place to simply store XML.  It is the relationship between the
documents that is often important (and essential) to the success of the
application.  Without an (indexed) database, these relationships are very
difficult to maintain.  The database itself does not (or should not)
perform any serialization.  A database that stores XML in its native form
is ideal, and this is realized in quite a few XML databases.

To address your concern, I think that if developers work backwards from a
descriptive XML schema to a language-specific object, then the pitful of
settling for the (non-descriptive) default serialization is avoided.  Then
the descriptive power of XML is utilized to its fullest.  However, I have
to admit, when designing a new application for which there are no
predefined XML schema's, I often do settle for the default mapping.  It is
the simplest and fastest solution.

-Patrick

Joe Landman <landman at scientificappliance.com>@bioinformatics.org on
07/11/2002 09:08:33 AM

Please respond to biodevelopers at bioinformatics.org

Sent by:    biodevelopers-admin at bioinformatics.org

To:    biodevelopers <biodevelopers at bioinformatics.org>
cc:

Subject:    RE: [Biodevelopers] what databases are you using?

I agree on the need for flexible XML databases.

What has worried me is the serialization of objects.  The beauty of XML
is of course that it is a self descripted structured datatype.  Building
objects using XML to represent the data structures is a powerful
technique (been doing that with Perl for more than a year).  The problem
is storing the object in a DB.

Serialization has always bugged me as non-optimal approach...  using a
database as a file system rather than a database.  Are there any
non-serializing DB's for XML out there?  It doesnt seem like a hard
problem to solve...

Aside from that, the utility of the DB is significantly lowered as a
language independent persistant data object repository if you need to
deal explicitly with one languages serialization methods (e.g. no
language independent de-serialization methods).  This point has also
irked me, and I dont know of a good solution to serialization (apart
from not serializing).  Add to this the performance hit of walking a
complex data structure and formatting it for the serialization
process...

Joe

On Thu, 2002-07-11 at 08:57, Patrick McConnell wrote:
>
> For many applications, we are using MySQL.  It has proven to be quite
> dependable and efficient.
>
> For more complex data (e.g. blast results), we are moving towards using
an
> XML database.  The free XML databases (Xindice, and a couple others) do
not
> perform particularly well, and we found them to be undependable, often
> hanging with frequent and simultaneous access.  Thus, we have invested in
a
> commercial product (Tamino), and have found it to be excellent.  Though
we
> are still in the development phase of our application, Tamino has worked
> great.
>
> I find the flexibility of XML to be essential to storing the often
complex
> biological data.  Also, several packages exist to automatically serialize
> and deserialize Java objects to XML.  This has greatly sped development
> time of our applications, as extremely complex data structures can be
> mapped directly to XML and indexed in a database.
>
> -Patrick McConnell
> Duke Bioinformatics Shared Resource
> mccon012 at mc.duke.edu
>
>
>
>
>
> "Pawel Krasucki" <Pkrasucki at psstat.com>@bioinformatics.org on 07/11/2002
> 08:34:19 AM
>
> Please respond to biodevelopers at bioinformatics.org
>
> Sent by:    biodevelopers-admin at bioinformatics.org
>
>
> To:    <biodevelopers at bioinformatics.org>
> cc:
>
> Subject:    RE: [Biodevelopers] what databases are you using?
>
>
> I was going to setup a test environment and was trying to figure what db
> to setup as well.  You mentioned using Postgres, have you ever worked
> with MySQL?  Although I've installed Postgres, I haven't worked with it
> at all, is it hard to pick up and work with?  I had it running on Linux,
> and was using GUI utility with Linux to administer Postgres,  is that a
> common way to administer that db?
>
> Paul Krasucki
> IT Supervisor
> Professional Solutions STAT Inc.,
> St. Louis, MO  63114
> (314) 428-8335 ext 129
>
>
> -----Original Message-----
> From: biodevelopers-admin at bioinformatics.org
> [mailto:biodevelopers-admin at bioinformatics.org] On Behalf Of Joe Landman
> Sent: Wednesday, July 10, 2002 09:55 PM
> To: biodevelopers
> Subject: [Biodevelopers] what databases are you using?
>
> A few years ago, the db of choice was some sort of Oracle implementation
> for many people I spoke to/worked with at a variety of pharmas and
> biotechs.  I am curious what people are using now (if anything).  I do
> know people using flat files for a variety of reasons as well.
>
> Just curious.  I use/like Postgres for my own development/testing.
>
> --
> Joe Landman,
> email: landman at scientificappliance.com
> web  : http://scientificappliance.com
>
> _______________________________________________
> Biodevelopers mailing list
> Biodevelopers at bioinformatics.org
> http://bioinformatics.org/mailman/listinfo/biodevelopers
>
> _______________________________________________
> Biodevelopers mailing list
> Biodevelopers at bioinformatics.org
> http://bioinformatics.org/mailman/listinfo/biodevelopers
>
>
>
> _______________________________________________
> Biodevelopers mailing list
> Biodevelopers at bioinformatics.org
> http://bioinformatics.org/mailman/listinfo/biodevelopers
--
Joe Landman,
email: landman at scientificappliance.com
web  : http://scientificappliance.com

_______________________________________________
Biodevelopers mailing list
Biodevelopers at bioinformatics.org
http://bioinformatics.org/mailman/listinfo/biodevelopers