Books/Accounting Periods proposal

[Christopher Browne]

On 09 Apr 2001 11:23:11 EDT, the world broke into rejoicing as
Derek Atkins <warlord@MIT.EDU>  said:
> Christopher Browne <cbbrowne@localhost.brownes.org> writes:
> > Consider the following control file:

> And how does one get this control file from client to client, user
> to user?

It obviously has to get copied somehow.  For a complex configuration,
it can be considered a given that configuration won't do itself.

In a multiuser environment, the administrator might stick the file
into /etc/skels, or put it on an NFS mount.

For a salesman with a laptop, the issues are largely analagous to
dealing with:
 -> Setting up email access on the road;
 -> Syncing spreadsheet and word processor documents between desktop 
    and laptop

In that case, copying the control file over seems in order, and seems
to be a not daunting task.

> What happens if different users have different control files on
> their clients while trying to access the same book?

If one user is trying to access an engine via the RPC interface,
another via CORBA, and another via talking straight to PostgreSQL,
there's _no_ issue, so long as the engine is capable of coping with
multiple users.

Frankly, if the engine is capable of coping with multiple concurrent
users, I don't see there being an issue.

A _thought_ is that the present XML file-based scheme represents an
approach that is inherently _NOT_ multiuser.  Thus, stick a lock file
in, and the second user simply can't get in.

To go to multiuser requires having either a storage scheme that does
_not_ involve the present serial access [as with serialization of
XML], but rather some form of multi-user-capable DB.  Or for there to
be a multithreaded engine, which gets accessed via RPC or CORBA or
such.

> However, I do like where this is heading.
--
(concatenate 'string "cbbrowne" "@acm.org")
http://vip.hyperusa.com/~cbbrowne/resume.html
"If you were to implement this in  ML, most of you would say : SCREAM,
run from the room in terror, but we could set it as a tickable problem
whatever, and  give it  to part 1A  [first year students]...  and that
idea seems rather pleasing doesn't it?"  -- Arthur Norman