Freechain question

[Kenneth Brody]

Quoting Don Coleman (Wed, 25 Jul 2007 14:36:45 -0400):

> > From: Kenneth Brody [mailto:kenbrody at bestweb.net]

[... Cancelled freechain while rebuilding. Restarted *clerk w/o issue ...]

> > Perhaps the corruption was near the beginning of the file, and it only
> > took a few seconds to reach that part and correct it?
>
> Possible, I guess.  However, the freechain rebuild was still running
> when I closed the app.  If it thought it had fixed any damaged portions
> of the freechain and sensed no more, wouldn't it have stopped (exited)
> the freechain rebuild?  Or would it continue to run, looking/rebuilding
> the entire freechain?  I'm not questioning you, I'm just trying to
> understand better what is happening/ triggering this, in the hopes of
> resolving the issue.

It rebuilds the entire freechain, requiring that it read the entire
file.  (It is not a simple "fix the first thing it finds wrong".)

[...]
> If the "auto" freechain rebuild is terminated by closing the window,
> what are the ramifications?

The freechain may still be corrupt.

Also, if "the window" is a telnet connection to a *nix box, you also
have the issue that some combinations of client/daemon/OS will not
shut down the telnet session that's running on the server.

> A) Another freechain rebuild will be triggered shortly, because the
> original cause has not been fixed.

That's one possibility.

> B) Any process adding records to this file will possibly take new
> records instead of using a previously used, now deleted record.

This only happens if the head of the freechain (in record zero) says
there are no free records.  This won't happen here unless the head
already was marked as such.

> Lastly, is the Windows version of filePro more prone to freechain
> rebuilds than the *unix versions?

Not that I am aware of.  Both the Windows and *nix versions of
filePro handle the freechain exactly the same way.  So, unless you
were to run into a bug in Windows or your network which were to
somehow cause such errors, the treatment of the freechain should
be identical.

> Or, because this is a 5.0.13 version, were older versions more prone
> to freechain rebuilds?

I don't recall any freechain bugs for many years.

> The main reason for not upgrading is my app. runs somewhat slower
> than I'd like currently, and I'm afraid introduction of the license
> manager used in the current version will only slow it further.

The license manager only causes a tiny delay in the startup of any
given module as it checks the license, or the very first time that
it attempts to access a separately-licensed feature.  Once it has
been granted the license, there is no performance hit for the
license manager.

> I realize these are somewhat risky questions to be asking on an
> open forum such as this, but I need to make some progress on
> several persistent issues.

--
KenBrody at BestWeb dot net        spamtrap: <g8ymh8uf001 at sneakemail.com>
http://www.hvcomputer.com
http://www.fileProPlus.com