[AccessD] OT: open source

 One thought on open source. Without a certain critical mass of users, 
support for certain types of open source software
can be a definite problem. This happens when you go beyond the common 
spreadsheet and word processing programs.

Let use an example. The two major general statistical packages SPSS and 
SAS do not run on Linux. They are used throughout government and large 
companies like pharmaceutical industries. Both these packages are taught 
in universities in courses like the Social Sciences. There is a large 
user support database, user conventions, mailing lists etc. The users 
have a familiarity with the packages.

 Now there is an open source replacement called R. Works on Unix, Linux 
and Windows.

R is possibly as good as SAS.  Probably better graphics than SAS-Graph. 
Cutting edge statistics - Many are developed in R (or in its commercial
cousin - S Plus).  It takes a while for them to make it into SAS. R 
doesn't have the data management abilities that SAS has.
However support is limited. The R language used is as different from SAS 
as English is to Swahili. So the ordinary social scientist doing 
regression analysis is going to have problem switching, whereas a recent 
Ph.D. in Statistics has probably been exposed to both languages. What I 
have seen in some large government departments is  a reasonable 
compromise using Base SAS for data management and exploratory data 
analysis and R for more advanced stats and graphs.

 Until recently R was restricted by only working with Data Sets in 
memory and access to external data was limited. In another year, they 
will be able to get at XML data, SAS  transport (XPORT) format data sets 
and some databases. It can get at Postgres now, higher versions of SQL 
and even Access.
 If you switch from SAS to R, you will have to overcome training and 
support problems  Is it worth the $1000 a seat difference? There are not 
a lot of people familiar with R (not enough critical mass of users or 
support personnel) . If some of the major pharmaceuticals switch to R, 
then the number of support personnel will increase. They will then endow 
some more universities to start teaching it to ensure a supply of 
trained personnel.

If anyone is interested in running Access and R on windows here is a 
starting point. I suppose you can also do it on Linux with Samba.

http://www.r-project.org/
http://www.stats.ox.ac.uk/pub/R/rw-FAQ.html

 Here is where a lot of open source packages fall down on data 
mangement. For example MAXDB (old SAPDB) from MySQL has a limited  ODBC 
driver and no OLEDB driver Would be nice to have to link to dotNet. 
However you will have to wait for some kind spirited soul to write one, 
It would take a least 3 months of someone's time. I have heard of people 
offering to do it for $10,000. Maybe someone will do it at MySQL. Guys 
who can write this type of code are thin on the ground.

This also happens with other large scale systems like AutoCAD or GIS 
systems like ArcINFO. ArcINFO can handle spatial database engines on 
Linux, but the front end Linux support requires Java Object interfaces, 
where there is little open source generic front ends available for 
datamangement. Although there are plenty of  frontend viewers available 
even through html and maybe svg xml. There are no decent free/open 
source CAD packages. Why? Because it is so damn difficult and 
time/manpower consuming to program one. Therefore to try and duplicate 
the functionality of any CAD package and especially the new generation 
parameteric modellers such as AutoCad Inventor, ProE, SolidWorks, etc. 
is not practical. I would say it is impossible with a limited 2D 
programs  such as open source MetaPost.

And come to think of it. I haven't seen any good open source accounting 
packages for Linux.

-- 
Marty Connelly
Victoria, B.C.
Canada