On Friday, May 23, 2003, at 02:33 AM, Kaleb wrote:
For example, one should be able to set up an adopath for a web site
that
contains a central set of personal/custom ados. You could then run
this
from anywhere in the world and know that you are always using the most
current versions. You should be able to use the "which" command with
this
as well (another command that doesn't work with http probably due to
the
same reason).
This is exactly the capability provided by Stata's "net" command, with
the advantage that one is not forced to subsidize Bill Gates to use it.
Now, I'm not your average user, but I do see the long-term benefits for
others of making this possible. Currently & locally, we keep our ados,
data, and profile.do's on a central server so that we don't have to
constantly update each computer individually. Now that we have 15+
users
across the world, we need something similar. I have created a useful
ado
archive system (easier to update/maintain than the .toc & .pck system
especially since most files are customized for our data) so these
users can
get the newer or older versions of any of our ado files; they are all
on
the web, and so far with XP it appears to work well.
The SSC archive runs with an 'ado archive system'--the .pkg files are
autogenerated/updated from the RePEc templates which describe each
archive entry, and the .toc files are automatically updated for new
entries when a change is made. The Distribution-Date is also modified
in that process. This is done with some fairly small and simple perl
routines. There is no need to abandon the flexibility of Stata's "net
from" tools in order to create customized ado-file collections (and
since a package can contain 'ancillary files' such as data files...) I
can readily imagine a true archive where the prior version of a routine
would be moved to a separate directory, allowing version tracking a la
CVS.
It is for the
Win2k/Mac users that this becomes a problem.
I think that anything developed for (world-wide-)workgroup use using
Stata must be done with platform independence in mind. Taking advantage
of OS-specific features is very inimical to the nature of today's
computing environment. It is not just non-updated Windozers and Mac OS
X users--what about the increasing number of Linux users out there, or
those (like myself) that make heavy use of Unix? Those data sites that
assume that everyone in the world uses Windows are doing a great
disservice to their potential clients (just as those dataset providers
who assume that everyone has Some Alternative Software running on their
desk...)
I also imagine that this will
be expanded to be a source of data files as well - it may be that we
will
just need to map the web site(s) to local drives for security and ease
of
use, but I thought you should at least know why this has become an
issue
for us.
Also, I am still curious as to what Stata does to check the existence
of
these web files checked using the -net- commands if -confirm file-
isn't used.
I imagine someone in the know @ StataCorp will answer that.
Thanks
Kit
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
