Re: diglib: database software

[JQ Johnson <jqj@darkwing.uoregon.edu>]

Bob's replies are right on.  My replies to his:

Apache on windows:  I don't think this a viable option yet.  I find apache 
on Linux to be excellent (and like Bob I know it well so am very 
comfortable with it), but apache on windows is a very immature product with 
many problems and without installed base.  For me, the point of running a 
web server on windows would be to get an environment that was "mainstream" 
and vendor supported.  That means IIS, I'm afraid.  As Bob notes "The only 
thing that excites me is something that is non-exciting."  On a more 
detailed technical level, I'd worry about lots of things with apache on 
windows: performance (IIS is reputed to substantially outperform Apache at 
this time, though that may eventually change), compatibility with ASP, 
compatibility with Windows 2000 authentication and active directory, 
security, maybe even compatibility with ColdFusion.
I spoke with Travis, and he does not see any problem in administering an 
IIS server.  But, as Bob notes, there's also the user training issue.

On ColdFusion:  Colleen made an excellent point to me.  She does not 
believe that anyone in the library, with the possible exceptions of Bob, 
Dave McCallum from NMC, and me, has the programming expertise+time to do 
significant PHP development.  In particular she notes that although she has 
developed 2 simple web sites that use PHP she did it by extensive borrowing 
from similar sites, and doesn't feel comfortable with going beyond this 
despite her programming background.  So if we were to go the PHP route it 
is pretty certain that our web development activities would become even 
more centralized than they are now, with Systems (or someone) contracting 
with Library departments to do the devel work.  Colleen believes that a 
decent development environment for ColdFusion would allow more distributed 
development.  I'm not sure, and it's the big issue that I'm currently 
anxious to explore:  can we expect a competent web author who is not a 
programmer/database developer to build a dynamic web site using DW 
Ultradev+ColdFusion+Access?  If not, then we're stuck with a 
centralized/contract model no matter what technology we choose, which 
paradoxically makes our choice less constrained.

The other issue implicit in Bob's note is a question of time frame.  Any 
technology in the web world has a finite and fairly short lifetime (for 
instance, we already need to revamp the static web pages on libweb due to 
changes in HTML standards and aesthetics).  But we'd like to get as long a 
lifetime as we can, at least let's say 2 years.  I'm uncertain at the 
moment about the near-term future of ColdFusion due to the 
Allaire/Macromedia merger and to Allaire's previous statements about 
wanting to convert ColdFusion to a JSP application server platform.

As I noted earlier, I think NMC futures are key here, since that's the 
piece of our organization that is likely to do most of the database-driven 
web development.