FW: Processing Password Questions

[Fairlight]

With neither thought nor caution, Jeff Harrison blurted:
> [snip]
> I agree that random doesn't usually happen :-) 
> However, things in the computing world often appear to
> be random.  For example the case in point that started
> this thread.

One of those things that separate the professionals in the field from the
hobbyists is -knowing- that even if it seems random, there's no way it
really is.  Sporadic, sure.  Unexpected, sudden, sure.  But there's always
a reason that should be found.

> While It has never happened to me, I can see that it
> would be easy - especially for someone not overly
> familiar with the PFPATH file - for someone to enter a
> fourth line without realizing exactly what they are
> doing - perhaps they are trying to get filepro to look
> in two possible places for its data.  Or perhaps they
> just enter some text there as a comment.  

Whether passwords are useful or not is a moot point when addressing this
though.  The question is begged:  If they don't know what they're doing,
and/or are unfamiliar with editing the file, why are they dinking with it
manually to begin with?  This speaks to needing a system administrator that
does know what they're doing, or needing to be more thorough in reading the
documentation.  This is the same gripe I have with people that run linux
and complain x, y, and z are broken; these people don't have the knowledge
to be going it alone in that environment, and then have the brass ones to
complain when the system acts "oddly" (at best, sometimes).

Education is key.  Either hire someone educated enough to do it correctly,
or educate yourself enough to do it correctly.  But you just don't haul off
and make changes to -any- subsystem you don't understand enough to
competantly administer.

Change all instances of "you" to "one" and change the corresponding tenses
and the like in the above.  Obviously -you- don't have problems, and you're
also clued.  I'm using sloppy grammar while referring to people that create
their own problems.  I have to wonder if some people just ram a car into
second gear without using the clutch, and later the same afternoon at the
mechanic's, wonder aloud why they have to pay almost a thousand bucks to
get their transmission fixed.  Ignorance is no excuse in any endeavor.

It's really no coincidence that you can tell those that are responsible for
-maintaining- stable systems from -using- stable systems by whether they
tend toward a preponderance of knowledgeable answers, or just lots and lots
of questions that are usually documented in TFM in the first place.

> I agree, though.  For ME site passwords are usually
> not worth the trouble.  I do use them though as a way
> to give a little bit of protection to my JHExport and
> JHimport demos.   

On that almost unrelated note, I just thought of the neatest thing...  If
you get the upcoming 5.6, you could theoretically encrypt an entire table
and have another table decrypt it based on a passphrase you could put in a
flat text file somewhere.  Decrypt the table, CALL it, and (I hope) have
the first thing the table does be to unlink itself so it can't be read
without a huge race condition.  There are a few details I probably haven't
thought of here, and it would be better if there was support for named
pipes when using processing tables (actually, that's a cool idea), but as a
totally instant, off-the-cuff concept, it's kind of interesting in terms of
protection.  You could issue a passphrase and a table encrypted with that
passphrase automatically, via a web application.  The details need some
work, but still...interesting thing you got me thinking about.

Actually, I wonder if it's possible after all.  I just got to thinking
that it would be a FIFO issue.  First in removes it, but might not be the
first one to be done needing to read it.  There could be race conditions
where it may fail in multiuser environments.  If it could be worked around
though...neat.  Maybe use EXISTS with a temp lockfile to prevent racing.

*ponder*  I'll have to give this more thought.  Impossible to even test
until 5.6 is actually released and one of my clients upgrades anyway.

mark->
-- 
Fairlight->   ||| "They beat him up until the        | Fairlight Consulting
  __/\__      ||| teardrops start / but he can't be  |
 <__<>__>     ||| wounded 'cos he's got no heart..." | http://www.fairlite.com
    \/        ||| --Elvis Costello                   | info at fairlite.com