I know that my experience is hardly an iron clad guarantee, but I have
to interject here that I have been using DateBk6 since its first beta,
and I think I can say just about the same for TealScan ...

I have *never* (and I sincerely mean "never", I'm not just trying to
over state things) seen either DateBk6 or its resource database show
up as listed as unlocked and/or unprotected.

I have used, in addition to TealScan, RescoLocker, PalmInsider, Palm
Internals, and a couple other analysis apps (whose names escape me
right now) for this sort of thing, many times, both in testing
TealScan and in sorting out my own NVFS problems on my 700p.

So, I can not help but think there is something else happening here.
I have no clue as to what ... but something.

Ken

P.S. ... and I use just about every feature of DateBk, except time
zones ...

--- In tealtalk@yahoogroups.com, Ralph Alvy <ralph@...> wrote:
>
> Thank you so much for this reply. It really helps a lot. I have
since locked
> Datebk6Resources with TealScan instead of RescoLocker and will see what
> happens. It still seems to be the only app that needs this attention
so far,
> after thoroughly exploring all the options you outlined below.
>
> On Fri May 23 2008 12:17:15 pm tealtex wrote:
> > In analyzing what's happening, don't fall into the trap of assuming
> > that any app is necessarily 100% accurate.  The safest thing to do is
> > turn off or lock everything that is suspect and see if crashes go
> > away.
> >
> > Also, it seems like you might be confusing some of the terms, so let
> > me try to clarify things so you don't jump to the wrong conclusion:
> >
> > 1) When applications register for system notifications, they can do
> > so thru an optional callback function.  Manually scan for these in
> > TealScan using TealScan's "BGProc" tool.  Any apps listed that also
> > have "(Callback)" next to them are a potential cause for random NVFS-
> > related crashes because they are passing PalmOS a function pointer to
> > a block of memory that cannot move.
> >
> > 2) When programs register for direct system notifications, they
> > typically pass PalmOS a pointer to code that resides in one of an
> > application's code resource records.  If so, that specific resource
> > record MUST be locked to keep the system from being unstable.
> > There's no way to tell externally in which record of a file that the
> > critical code resides, so any app tells you only that a file's code
> > is either totally "locked" or "unlocked" is misleading, and I
> > wouldn't rely on it.
> >
> > Instead, TealScan shows you how many code resources in a file are
> > locked compared to the total number of code resources.  If only some
> > or none of the code resources are locked, the file may or may not be
> > stable in this regard and you should probably manually lock it in
> > TealScan to insure all the code resources are in fact locked.
> >
> > 3) In addition to the code resource lock count, each file in memory
> > also has a file-based global "protect" count.  In early NVFS devices
> > like the T5 or Treo 650, locking code resource was enough to keep the
> > code in the DBCache, but on newer NVFS devices, NVFS is more
> > aggresive in flushing the DBCache and will flush locked records if
> > their owning file is not also "protected."
> >
> > When a program refers to a *file* being "locked", they are probably
> > only referring to the protect count.  You need to remember that both
> > the file and the resource record that holds the callback code must be
> > protected/locked.  Only doing one is insufficient.
> >
> > 4) Your test results are not as contractory as they may seem.
> > Remember, TealScan also checks the system notifications list for
> > Background processes, something that the other apps cannot do.
> >
> > TealScan won't list a file as a problem if it currently isn't
> > registered for notifications, as it doesn't really need to be
> > locked/protected at that time.  If the DateBk6Resources file appears
> > and disappears between scans, the file/records may either be
> > unprotected/unlocked some of the time, or maybe instead DateBk6 does
> > not always keep the code in that file registered for notifications
> > all the time.
> >
> > To check whether individual files/records are actually
> > protected/locked, use TealScan's "locked applications" menu instead.
> > You'll probably get results that match RescoLocker.  If SafeGuard
> > doesn't distinguish between locked records and protected files, I
> > would not rely on it.
> >
> > It sounds like the DateBk6Resources file is only occassionally
> > registered for notifications, but sometimes is either registered
> > without proper locking/protecting, or alternatively is
> > unlocked/unprotected by DateBk6 without being unregistered for
> > notifications first.
> >
> > If either of these is true, then yes, you should try to manually lock
> > the file/code resources with RescoLocker or TealScan.  You need to
> > still monitor the files, however, as DateBk6 has the ability to
> > remove any locks or protection that you put on externally, and it
> > could do that at any time.
> >
> > -tex
> >
> > --- In tealtalk@yahoogroups.com, Ralph Alvy <ralph@> wrote:
> > > The only unprotected BG app that TealScan occasionally shows is
> > > Datebk6Resources. Never anything else. And most of the time, it
> >
> > never shows
> >
> > > any unprotected BG apps at all.
> > >
> > > RescoLock usually shows Datebk6 unlocked and unprotected. Sometimes
> >
> > it shows
> >
> > > Datebk6Resources that way as well. Right now it shows Datebk6
> >
> > unprotected and
> >
> > > unlocked. Safeguard always shows both Datebk6 and Datebk6Resources
> >
> > as safely
> >
> > > locked.
> > >
> > > Perhaps I should have RescoLock go ahead and lock Datebk6Resources
> >
> > and see
> >
> > > what happens. What do you think?
> > >
> > > On Thursday 22 May 2008 09:42:46 pm tealtex wrote:
> > > > Again, I don't think this crash has anything to do with a bug in
> > > > either TealLock or any of the backup apps you are running.
> >
> > SafeGuard
> >
> > > > shows TealLock because it is the first app launched after the
> >
> > backup,
> >
> > > > and the backups cause a crash because the backup process runs
> >
> > through
> >
> > > > all the files in memory, which cause the NVFS system to move files
> > > > around in memory.
> > > >
> > > > Our experience with TealBackup has shown us that if any unlocked
> > > > background apps are running, doing a full backup will cause them
> >
> > to
> >
> > > > crash, and the crash will often occur the next time an app enters
> >
> > an
> >
> > > > interactive event loop.  This sounds exactly what you are seeing
> >
> > on
> >
> > > > your device.
> > > >
> > > > Flushing or not flushing the cache will not do anything, as the
> > > > background apps will already have been moved during the backup
> > > > process, nor will changing any settings in TealLock.  Please try
> > > > using TealScan to look for any unlocked/unprotected background
> >
> > apps.
> >
> > > > -tex
> > > >
> > > > --- In tealtalk@yahoogroups.com, Ralph Alvy <ralph@> wrote:
> > > > > I still occasionally get this reset. Got it again last night.
> >
> > The
> >
> > > > pattern is
> > > >
> > > > > fairly consistent. While TealLock has my Centro locked,
> >
> > BackupMan
> >
> > > > (9:30pm) or
> > > >
> > > > > RescoBackupPro (10:00pm) does a Full Backup, flushes the cache,
> >
> > and
> >
> > > > exits,
> > > >
> > > > > all while locked by TealLock. If it's going to reset, it will
> >
> > do so
> >
> > > > when
> > > >
> > > > > TealLock takes back control.
> > > > >
> > > > > I have tried having these apps on the Allowed list, and off the
> > > >
> > > > Allowed list.
> > > >
> > > > > I have tried disabling their flushing of the cache. No changes.
> > > >
> > > > Still get the
> > > >
> > > > > occasional resets.
> > > > >
> > > > > Since I have yet to see this happen after an Update Backup, I'll
> > > >
> > > > turn off all
> > > >
> > > > > Full Backups for RescoBackupPro and see what happens with only
> > > >
> > > > Update Backups
> > > >
> > > > > performed at 10:00pm.
> > > > >
> > > > > On Thursday 22 May 2008 03:22:54 pm tealtex wrote:
> > > > > > Now knowing how SafeGuard works internally, I would caution
> > > >
> > > > against
> > > >
> > > > > > relying absolutely on what app it reports.
> > > > > >
> > > > > > An illegal instruction is usually caused by a background app
> >
> > that
> >
> > > > does
> > > >
> > > > > > not lock itself down properly, so PalmOS tries to execute
> >
> > random
> >
> > > > bytes
> > > >
> > > > > > after the NVFS system has moved the app.  TealLock does lock
> >
> > its
> >
> > > > code
> > > >
> > > > > > resources, however, which you can verify with TealScan when
> > > >
> > > > TealLock
> > > >
> > > > > > is "ON".
> > > > > >
> > > > > > SafeGuard listed TealLock as the last app "run", but I suspect
> > > >
> > > > this
> > > >
> > > > > > means that the last app launched with a launch code was
> > > >
> > > > TealLock.  Apps
> > > >
> > > > > > that register for notifications thru a direct callback pointer
> > > >
> > > > aren't
> > > >
> > > > > > ever "run" in this way, and these are exactly the types of
> >
> > apps
> >
> > > > that
> > > >
> > > > > > cause illegal instructions if they don't lock themselves down
> >
> > for
> >
> > > > NVFS.
> > > >
> > > > > > I would turn off all other background apps that you have
> >
> > running,
> >
> > > > > > and/or run TealScan to see if it can spot an obviously
> >
> > misbehaved
> >
> > > > app.
> > > >
> > > > > > -tex
>

Ken Latham wrote:
> I have *never* (and I sincerely mean "never", I'm not just trying to
> over state things) seen either DateBk6 or its resource database show
> up as listed as unlocked and/or unprotected.
  >
  > <... clipped text ...>
  >
> Ken
>
> P.S. ... and I use just about every feature of DateBk, except time
> zones ...


Well, last night, when I read your message, I checked RescoLocker. It
listed Datebk6 with an exclamation point all by itself, signifying it
receives notifications, but is unlocked and unprotected. It showed
Datebk6Resources as LP without an exclamation point, signifying it's
locked and protected, but not ready to receive notifications. The latter
is locked and protected because I have TealScan doing that.

This morning RescoLocker says Datebk6 is in the same state it was last
night. However, it now lists Datebk6Resources with !LP, so it is not
registered to receive notifications, while being locked and protected by
TealScan.

Ralph Alvy wrote:
>
>
> Ken Latham wrote:
>  > <... clipped text ...>
>  >
>  > Ken
>  >
>  > P.S. ... and I use just about every feature of DateBk, except time
>  > zones ...

I use TimeZones, if that's relevant.

I have been a Palm user forever, and love TealLock and Databk6.  What I
call the "destructive password" aspect of TealLock is crucial for me,
since I have sensitive data on my Palm.

I'm considering moving to a smartphone, and not a treo, and it doesn't
look like TealLock will work on that.  Does it?  Do smartphones have
the equivalent, either standard or for purchase?

I am in the healthcare field and have to be HIPAA compliant.  Does
TealPoint or anyone like them make data encryption programs for MS
Mobile 6?

Thanks for your help,

Deborah

I'm really not sure about this one.  When you set up phone dialing in
TealPhone, it's usually done through two methods.

The first method is to dial directly using AT-style modem commands to
any serial driver that has registered itself with palmOS.  When done
thru the bluetooth virtual serial driver, it can be used to connect to
a phone that both supports a serial-over-bluetooth type of connection
and is set up to emulate a serial data modem that supports Hayes AT-
stle commands.  Most GSM phones support some variation of this type of
connection, but to use this method you really need to know a lot about
the internals of the phone.

The second and more common method is to dial through the Palm Telephony
library using the "PalmOS" integrated driver.  On smartphones, this
directly initiates a call, while on other devices (like a TX) this
initiates a call through whatever external phone has been set up in
Palm "Phone" Preferences.  We suspect that this ultimately also
connects to the external phone via serial bluetooth, but automaticallys
adds support for different dialing commands.

When trying to dial to your Treo, setting up a bluetooth connection
itself is probably not enough.  You also need to load an appropriate
phone driver for the Treo 650 onto your TX, and configure your "Phone"
prefs in the Palm Preferences app to point to the Treo.

A quick search shows a link to Palm.com where a Treo650 driver can be
found:

http://kb.palm.com/SRVS/CGI-BIN/WEBCGI.EXE?
New,Kb=EUPalmSupport,ts=PALMEU_UK,Case=Obj(7039),Question=obj(5356):obj
(7139)#1

Note, however, that only two US carriers (Cingular and T-Mobile) are
listed as being compatible, and the page implies that the drivers may
be intended only for "GSM" phones.  This might mean that non-GSM phones
do not typically support modem emulation, and/or that running as a
modem has purposefully been disabled by some carriers to discourage
using their network for laptop use without extra fees.

Along those lines, carriers like Sprint purposefully disable the built-
in dial-up-networking (DUN) preferences panel in Treos branded for
their network.  So if you have a non-GSM Treo or another carrier than
those listed on Palm's site, your mileage may vary.

-tex


--- In tealtalk@yahoogroups.com, "Dwight C. Carr" <carrdwight@...>
wrote:
> Another topic:  I have been trying to set up my TX to initiate calls
on my Treo 650 via Bluetooth.  Is this possible?  The devices are
paired, but the Treo will not dial the number that the TX sends.  It is
frustrating that I can initiate calls via bluetooth to an old Nokia
phone, but not to the Treo. 
> Thanks,
> Dwight...

Hi.

We don't have any immediate plans to port TealLock to Windows Mobile,
but a German company named Utimaco makes a line of powerful
enterprise-level security products called "SafeGuard xxx" (SafeGuard
Enterprise, SafeGuard PDA, SafeGuard PrivateCrypto, etc) that might
meet your needs.  SafeGuard PDA has been around for many years and is
unrelated to the crash-detection app that GoTreo released in March of
this year and confusingly named "SafeGuard for PalmOS".

SafeGuard PDA is available for Windows Mobile, Symbian, and PalmOS
devices.  The PalmOS version is especially similar to TealLock
because it is actually an OEM version of TealLock that we developed
for Utimaco:

http://americas.utimaco.com/safeguard_pda/download.html

-tex

--- In tealtalk@yahoogroups.com, "dtmfcc" <deborah1633@...> wrote:
>
> I have been a Palm user forever, and love TealLock and Databk6.
What I
> call the "destructive password" aspect of TealLock is crucial for
me,
> since I have sensitive data on my Palm.
>
> I'm considering moving to a smartphone, and not a treo, and it
doesn't
> look like TealLock will work on that.  Does it?  Do smartphones
have
> the equivalent, either standard or for purchase?
>
> I am in the healthcare field and have to be HIPAA compliant.  Does
> TealPoint or anyone like them make data encryption programs for MS
> Mobile 6?
>
> Thanks for your help,
>
> Deborah
>

--- In tealtalk@yahoogroups.com, Ralph Alvy <ralph@...> wrote:
>
> Ralph Alvy wrote:
> >
> >
> > Ken Latham wrote:
> >  > <... clipped text ...>
> >  >
> >  > Ken
> >  >
> >  > P.S. ... and I use just about every feature of DateBk, except time
> >  > zones ...
>
> I use TimeZones, if that's relevant.
>

Have you brought this up with CESD over in the DateBk group?  I don't
recall seeing anything.  He'd want to know if some feature is causing
a notification problem.

... so, yes, time zones may affect what notifications are used ... at
least I think it would (not a Palm programmer myself).

I reinstalled Rlocker just to see if I could catch DateBk6 in the act
... and still have not been able to do so yet!  I live and breath by
DateBk6, so I am very curious ...

Ken

Just a quick note. I use TealLock 6.87 and DB6 with no problem.

Another beautiful day in paradise.
-----Original Message-----
From: "Ken Latham" <clatham1@...>
Date: Wednesday, May 28, 2008 5:19 pm
Subject: [tealtalk] Re: TealLock 6.86 reset
To: tealtalk@...: tealtalk@yahoogroups.com



--- In tealtalk@yahoogroups.com, Ralph Alvy <ralph@...> wrote:
>
> Ralph Alvy wrote:
> >
> >
> > Ken Latham wrote:
> >  > <... clipped text ...>
> >  >
> >  > Ken
> >  >
> >  > P.S. ... and I use just about every feature of DateBk, except time
> >  > zones ...
>
> I use TimeZones, if that's relevant.
>

Have you brought this up with CESD over in the DateBk group?  I don't
recall seeing anything.  He'd want to know if some feature is causing
a notification problem.

.. so, yes, time zones may affect what notifications are used ... at
least I think it would (not a Palm programmer myself).

I reinstalled Rlocker just to see if I could catch DateBk6 in the act
.. and still have not been able to do so yet!  I live and breath by
DateBk6, so I am very curious ...

Ken

Just a quick note. I use TealLock

Another beautiful day in paradise.
-----Original Message-----
From: "Ken Latham" <clatham1@...>
Date: Wednesday, May 28, 2008 5:19 pm
Subject: [tealtalk] Re: TealLock 6.86 reset
To: tealtalk@...: tealtalk@yahoogroups.com



--- In tealtalk@yahoogroups.com, Ralph Alvy <ralph@...> wrote:
>
> Ralph Alvy wrote:
> >
> >
> > Ken Latham wrote:
> >  > <... clipped text ...>
> >  >
> >  > Ken
> >  >
> >  > P.S. ... and I use just about every feature of DateBk, except time
> >  > zones ...
>
> I use TimeZones, if that's relevant.
>

Have you brought this up with CESD over in the DateBk group?  I don't
recall seeing anything.  He'd want to know if some feature is causing
a notification problem.

.. so, yes, time zones may affect what notifications are used ... at
least I think it would (not a Palm programmer myself).

I reinstalled Rlocker just to see if I could catch DateBk6 in the act
.. and still have not been able to do so yet!  I live and breath by
DateBk6, so I am very curious ...

Ken

Does a program exist that will allow a Bluetooth Headset such is used
with cell phones to record on my TX, from the BT headset and play that
recording and MP3 files to the BT Headset ???

On Wednesday 28 May 2008 08:19:22 pm Ken Latham wrote:
> Have you brought this up with CESD over in the DateBk group?  I don't
> recall seeing anything.  He'd want to know if some feature is causing
> a notification problem.

I mentioned this to CESD in private email. He wasn't all that concerned about
it. I've had no problems since I let TealScan lock/protect Datebk6Resources,
though I also locked/protected other items as well when doing that. I'm
slowly releasing them, one at a time, waiting a few days to make sure there
are no resets before releasing each one from TealScan locking/protecting it.

I would not think that he would be too concerned about it ... He's
quite possibly the best versed Palm programmer on the planet.  So, I
just can't see him making a mistake in that area, especially after all
the trouble he had tracking down bugs in Palm's implementation when
NVFS first came out.

I have a few programs that do show up in RLock as a "problem" when in
fact there seems to be a way to avoid any NVFS problems and still not
lock in a way that RLock "sees" (according to the apps' programmers).

I went through hell and back around Oct. of last year studying this in
detail, just to get my Treo to stop crashing.  I finally succeeded (as
in 7 crashes in as many months since, each and every one accounted for).

So, just from a user level testing perspective alone, I can say that
RLock's annotations are advisory, not conclusive.  And if you ask, the
guys at Resco will tell you this themselves.  My machine is stable,
*with* these "bad" programs running every day.

Unfortunately, the only certain method of weeding out apps that do not
handle NVFS correctly, is to trial each app one at a time for a few
days.  Actually, I did it in "batches", then if I had a problem, I
backed that batch out and re-installed just those one at a time.

The real killers are these:
(1) A soft reset after uninstall is sometimes not sufficient to clear
out some apps.
(2) A misbehaving app often does damage, and while I'm not sure why,
it seemes to *frequently* be to the saved preferences.  Once this
damage is done, no number of soft or warm resets (or even restorations
from backup) will fix it.

So if you add an app, start having resets, back it out, soft/warm
reset and still have problems, you are in for a hard reset and manual
install to get rid of the damage.

Fortunately, damage to saved preferences is not all that frequent, and
sometimes even recognizable, so it may be cured by uninstalling the
app whose prefs were damaged, and reinstalling it.

Compared to changing the core PIM apps, this NVFS problem is by far
worse.  Stupidest (mostly through knowledge distribution mishandling)
thing Palm has ever done, in my opinion.  And if it was intentional
then it was a backward incompatibility problem worthy of Microsoft!

Ken

--- In tealtalk@yahoogroups.com, Ralph Alvy <ralph@...> wrote:
>
> On Wednesday 28 May 2008 08:19:22 pm Ken Latham wrote:
> > Have you brought this up with CESD over in the DateBk group?  I don't
> > recall seeing anything.  He'd want to know if some feature is causing
> > a notification problem.
>
> I mentioned this to CESD in private email. He wasn't all that
concerned about
> it. I've had no problems since I let TealScan lock/protect
Datebk6Resources,
> though I also locked/protected other items as well when doing that. I'm
> slowly releasing them, one at a time, waiting a few days to make
sure there
> are no resets before releasing each one from TealScan
locking/protecting it.
>

On Thursday 29 May 2008 06:01:29 pm Ken Latham wrote:
> I would not think that he would be too concerned about it ... He's
> quite possibly the best versed Palm programmer on the planet.  So, I
> just can't see him making a mistake in that area, especially after all
> the trouble he had tracking down bugs in Palm's implementation when
> NVFS first came out.
>
> I have a few programs that do show up in RLock as a "problem" when in
> fact there seems to be a way to avoid any NVFS problems and still not
> lock in a way that RLock "sees" (according to the apps' programmers).

I never consider RLock the ultimate honcho to listen to. I rely on comparing
its results with TealScan and SafeGuard. The only app TealScan would complain
about was Datebk6Resources, as surprising as that sounds. RLock and SafeGuard
complained about a lot more than that, especially SafeGuard.

I also rely heavily on Datebk6, running my entire practice on it for years,
starting with Datebk5 (can't remember if I used DB4), and follow CESD in the
news.gmane.org group. I'm aware of his reliability in this regard.

> Compared to changing the core PIM apps, this NVFS problem is by far
> worse.  Stupidest (mostly through knowledge distribution mishandling)
> thing Palm has ever done, in my opinion.  And if it was intentional
> then it was a backward incompatibility problem worthy of Microsoft!

I'm assuming the NVFS was implemented in order to have nonvolatile flash
memory. I came to the Palm platform from the HP200LX, which always had
nonvolatile flash memory. All my data on that device was stored on a Compact
Flash card. I've never understood why this NVFS was so necessary to
accommodate nonvolatile memory, but I'm not expert in this area by any means,
as this thread can testify to.

There seems to be some confusion about how NVFS interacts with some
apps, so if I may, let me shed some light on why apps are sometimes
incompatible with NVFS.  I think this may help explain some of the
problems and proper ways to fix them.

Background on NVFS:

NVFS, as most of you know, was necessary to give PalmOS the ability
to retain data after a power loss.  However, flash memory is too slow
to be used as normal operating RAM.  While old devices had a big
block of RAM that constantly needed power, newer devices have a big
block of slow flash memory and a small pool of working memory known
as the DBCache.

Since there is more flash memory than real working memory, PalmOS has
to constantly shuttle chunks of files to and from flash to real
memory, making them available to apps when they need it, and saving
copies back to nonvolatile memory when they are done.

The problem is that many older apps were written to assume that once
they got a chunk (record) from a file that it would stay in the same
place.  This is largely true in older devices, but not on newer ones
where the chunk of memory they are given is in the temporary DBCache.

Crashes:

The primary cause of crashes occurs when an app gets a pointer to
some of its own code, and hands it to PalmOS as a callback function
for a given system notification.  If the apps doesn't explicitly lock
the chunk and protect the file, PalmOS will crash when it tries to
use the function pointer after the app has been flushed from the
cache.

In an ideal world, PalmOS would simply cross reference its internal
list of callback function pointers with any blocks it wants to move
or flush from the cache, but alas it does not do this.  So TealScan
was designed to warn you of this, reporting any files that have
direct callback functions registered for it but no locked code
records and/or an unprotected data file.

Other crashes and corruption, however, can occur if an app itself
decides to store a pointer internally, using it at a later time after
the block of memory it belonged to had been flushed.  When this
happens, it can overwrite other files or PalmOS itself, causing
spurious unpredictable crashes that vary depending on what files are
located in the cache and at what relative location.  These are the
most difficult types of crashes to track down, as the device will
appear to be stable for awhile until you do something completely
innocuous (such as installing a new file) that shifts around memory a
bit.

I wouldn't necessarily blame Palm entirely for this last problem, as
this is largely bad programming that can occur on any platform.
Their biggest mistake was not making available cache tools for
developers to flush the cache at will to help track down these sorts
of problems.

For the user, the only way to avoid these problems is to stick with
developers you trust.  If you suspect an NVFS-based problem, try
locking everything suspicious to see if the problem goes away.

App Installation:

Since the question was brought up, I also wanted to address the issue
of apps being installed or uninstalled.  PalmOS is almost entirely
file-based, so everything about the state of your device and the apps
installed on it are stored in files.

This means that when an app is installed, you can always totally
uninstall it by removing all of its files and undoing any changes it
may have made to files already on the device.  The only exception to
this is if (as described earlier) the installation process caused
problems because it happend to move files around, but then the
problem isn't caused by the newly installed app, but by something
that was already running.

The only system file that apps really modify is the Saved Preferences
database.

The trick is to know that when you "uninstall" an app using the
PalmOS launcher, it only removes files and Saved Preferences entries
that have the same CreatorID as the application.  Some apps stash
extra information (such as when an expiring trial version was
installed) by imbedding it in a separate file or in a Saved
Preferences records with a non-matching (orphan) CreatorID.  Other
apps do this purely because of bad programming.

Thus, with careful monitoring and backup of which files and
preferences are on your device, you can always manually uninstall any
app.

Use TealScan's "monitor saved preferences" option, as it will
automatically track any orphan CreatorIDs that an app creates.  It
also transmits these matches to our online database, so other users
can identify orphan preferences even if they install TealScan
afterwards.

-tex



--- In tealtalk@yahoogroups.com, "Ken Latham" <clatham1@...> wrote:
>
> I would not think that he would be too concerned about it ... He's
> quite possibly the best versed Palm programmer on the planet.  So, I
> just can't see him making a mistake in that area, especially after
all
> the trouble he had tracking down bugs in Palm's implementation when
> NVFS first came out.
>
> I have a few programs that do show up in RLock as a "problem" when
in
> fact there seems to be a way to avoid any NVFS problems and still
not
> lock in a way that RLock "sees" (according to the apps'
programmers).
>
> I went through hell and back around Oct. of last year studying this
in
> detail, just to get my Treo to stop crashing.  I finally succeeded
(as
> in 7 crashes in as many months since, each and every one accounted
for).
>
> So, just from a user level testing perspective alone, I can say that
> RLock's annotations are advisory, not conclusive.  And if you ask,
the
> guys at Resco will tell you this themselves.  My machine is stable,
> *with* these "bad" programs running every day.
>
> Unfortunately, the only certain method of weeding out apps that do
not
> handle NVFS correctly, is to trial each app one at a time for a few
> days.  Actually, I did it in "batches", then if I had a problem, I
> backed that batch out and re-installed just those one at a time.
>
> The real killers are these:
> (1) A soft reset after uninstall is sometimes not sufficient to
clear
> out some apps.
> (2) A misbehaving app often does damage, and while I'm not sure why,
> it seemes to *frequently* be to the saved preferences.  Once this
> damage is done, no number of soft or warm resets (or even
restorations
> from backup) will fix it.
>
> So if you add an app, start having resets, back it out, soft/warm
> reset and still have problems, you are in for a hard reset and
manual
> install to get rid of the damage.
>
> Fortunately, damage to saved preferences is not all that frequent,
and
> sometimes even recognizable, so it may be cured by uninstalling the
> app whose prefs were damaged, and reinstalling it.
>
> Compared to changing the core PIM apps, this NVFS problem is by far
> worse.  Stupidest (mostly through knowledge distribution
mishandling)
> thing Palm has ever done, in my opinion.  And if it was intentional
> then it was a backward incompatibility problem worthy of Microsoft!
>
> Ken
>
> --- In tealtalk@yahoogroups.com, Ralph Alvy <ralph@> wrote:
> >
> > On Wednesday 28 May 2008 08:19:22 pm Ken Latham wrote:
> > > Have you brought this up with CESD over in the DateBk group?  I
don't
> > > recall seeing anything.  He'd want to know if some feature is
causing
> > > a notification problem.
> >
> > I mentioned this to CESD in private email. He wasn't all that
> concerned about
> > it. I've had no problems since I let TealScan lock/protect
> Datebk6Resources,
> > though I also locked/protected other items as well when doing
that. I'm
> > slowly releasing them, one at a time, waiting a few days to make
> sure there
> > are no resets before releasing each one from TealScan
> locking/protecting it.
> >
>

See below ...

--- In tealtalk@yahoogroups.com, tealtex <no_reply@...> wrote:
>
> There seems to be some confusion about how NVFS interacts with some
> apps, so if I may, let me shed some light on why apps are sometimes
> incompatible with NVFS.  I think this may help explain some of the
> problems and proper ways to fix them.
>
> Background on NVFS:
>
> NVFS, as most of you know, was necessary to give PalmOS the ability
> to retain data after a power loss.  However, flash memory is too slow
> to be used as normal operating RAM.  While old devices had a big
> block of RAM that constantly needed power, newer devices have a big
> block of slow flash memory and a small pool of working memory known
> as the DBCache.
>
> Since there is more flash memory than real working memory, PalmOS has
> to constantly shuttle chunks of files to and from flash to real
> memory, making them available to apps when they need it, and saving
> copies back to nonvolatile memory when they are done.
>
> The problem is that many older apps were written to assume that once
> they got a chunk (record) from a file that it would stay in the same
> place.  This is largely true in older devices, but not on newer ones
> where the chunk of memory they are given is in the temporary DBCache.
<snipped a bunch>

For my edification ...

As I understand it, you (a programmer) can also "dynamically" lock
records, no?  By that I mean you ask the OS to lock the record only
just before you access it and then release it again once you have
written it ...  thus *most* of the time, the memory "chunk" would
appear to be unlocked to any app that went looking.

I understand this works because the OS will reload the chunk in
DBCache when the lock request is made, if it had already flushed it to
NVRAM.  Thus, if my understanding is correct, it would be a perfectly
safe way to program an app, and yet it would fool most of the lock
checkers into believing that the chunk is unprotected.

Ken

P.S. I still find it very curious that having tried many times in the
course of the last few days to catch it ... I have not seen DateBk6 or
it resource DB being flagged by RLock.  I still have this sneaking
suspicion that using time zones alone would not be the cause of
continually seeing an unlocked chunk ... but if CESD doesn't think its
a problem ... who am I to argue with him!?  Maybe there is something
DateBk6 must keep continually available in that case ...

P.P.S. (replying to a part I already snipped :/ ) ... It is not the
fact that they used NVRAM that I was objecting to (lord no), but
rather the sloppiness of the algorithms used to swap it.  Well
designed swap (paging, whatever) is entirely transparent (except
perhaps for what should be largely inperceptible delays).

Thus I find a paging algorithm, which is not only not automatic and
not transparent, an abomination to the name of OSs everywhere.  I have
an extremely difficult time accepting that it was not intentional.

And while you may find "abomination" a strong word, what with me not
being a Palm programmer, you may find that I know a good deal more
about that particular subject than you might guess.  Afterall, the
idiosyncracies of paging/swapping were worked out about 20 years ago.
  It should have been a dropkick regardless of what hardware they used.

Hi.

--- In tealtalk@yahoogroups.com, "Ken Latham" <clatham1@...> wrote:
> For my edification ...
>
> As I understand it, you (a programmer) can also "dynamically" lock
> records, no?  By that I mean you ask the OS to lock the record only
> just before you access it and then release it again once you have
> written it ...  thus *most* of the time, the memory "chunk" would
> appear to be unlocked to any app that went looking.

Nope.  The last thing an app must do before accessing a chunk of data
is lock the chunk, because the function that locks the chunk
(MemHandleLock) is also the same one returns a pointer to the chunk's
data.  The pointer is just a memory location inside the DBCache, so
the app has to keep the chunk locked to make sure NVFS doesn't swap
something else into that same space.

I think you're referring to the "DmGetRecord" and "DmQueryRecord"
calls apps use to get exclusive access when writing data to records.
This doesn't preclude the need to lock records, however, as the
actual functions used to read or write data still require a pointer,
not a handle, which is what DmGetRecord returns.  System notification
callback functions are also pointers.

> I understand this works because the OS will reload the chunk in
> DBCache when the lock request is made, if it had already flushed it
to
> NVRAM.  Thus, if my understanding is correct, it would be a
perfectly
> safe way to program an app, and yet it would fool most of the lock
> checkers into believing that the chunk is unprotected.

Yes, but an app that does what you describe (locks any data it needs
before use) won't crash when the DBCache is flushed doesn't need to
be protected in the first place.  So it wouldn't be "fooling" any
lock checkers because, while it wouldn't be locked/protected, it
wouldn't need to be either.

A problem exists only if an app lists "Callback" next to its name in
the TealScan BGProc list, as that means it has given PalmOS a
POINTER, and pointers are only valid while locked.  While a couple of
apps (like TakePhone) make copies of their callback functions into
dynamic RAM and lock them there, most apps lock them in place in the
program PRC file.  This is why the combination of a "callback" entry
in the BGProc list along with unlocked/unprotected code resources
usually indicates a problem.

> P.S. I still find it very curious that having tried many times in
the
> course of the last few days to catch it ... I have not seen DateBk6
or
> it resource DB being flagged by RLock.  I still have this sneaking
The next time it happens, manually view the BGProc list in TealScan
to see what the specific notification code is.  This might help
figure out what's going on.

> P.P.S. (replying to a part I already snipped :/ ) ... It is not the
> fact that they used NVRAM that I was objecting to (lord no), but
> rather the sloppiness of the algorithms used to swap it.

Ever tried a WM device?  Our 624Mhz WM handheld is still painfully
slow compared to my Centro.  While I think NVFS could have been much
better designed in terms of stability and availability of diagnostic
tools (like, how about a way manually shuffle the DBCache or examine
its contents!), I don't see much of a slowdown except when launching
Blazer, and I'm not sure that is entirely NVFS-related.

-tex

>
> Ken
>
> P.S. I still find it very curious that having tried many times in
the
> course of the last few days to catch it ... I have not seen DateBk6
or
> it resource DB being flagged by RLock.  I still have this sneaking
> suspicion that using time zones alone would not be the cause of
> continually seeing an unlocked chunk ... but if CESD doesn't think
its
> a problem ... who am I to argue with him!?  Maybe there is something
> DateBk6 must keep continually available in that case ...
>
> P.P.S. (replying to a part I already snipped :/ ) ... It is not the
> fact that they used NVRAM that I was objecting to (lord no), but
> rather the sloppiness of the algorithms used to swap it.  Well
> designed swap (paging, whatever) is entirely transparent (except
> perhaps for what should be largely inperceptible delays).
>
> Thus I find a paging algorithm, which is not only not automatic and
> not transparent, an abomination to the name of OSs everywhere.  I
have
> an extremely difficult time accepting that it was not intentional.
>
> And while you may find "abomination" a strong word, what with me not
> being a Palm programmer, you may find that I know a good deal more
> about that particular subject than you might guess.  Afterall, the
> idiosyncracies of paging/swapping were worked out about 20 years
ago.
>  It should have been a dropkick regardless of what hardware they
used.
>

--- In tealtalk@yahoogroups.com, tealtex <no_reply@...> wrote:
<snipped> ... This is why the combination of a "callback" entry
> in the BGProc list along with unlocked/unprotected code resources
> usually indicates a problem.

Thanks for the detailed explaination.  Probably boring to tears the
rest of the readers here ... ;)

>
> > P.S. I still find it very curious that having tried many times in
> the
> > course of the last few days to catch it ... I have not seen DateBk6
> or
> > it resource DB being flagged by RLock.  I still have this sneaking
> The next time it happens, manually view the BGProc list in TealScan
> to see what the specific notification code is.  This might help
> figure out what's going on.
>

Maybe I should not have changed gears there so quickly.  That was
referring to the other poster who was seeing them listed ... I never
have ... not once.  That is why I was bringing it up.  It seemed odd
that they should have them show all the time, and yet, I never do!


> > P.P.S. (replying to a part I already snipped :/ ) ... It is not the
> > fact that they used NVRAM that I was objecting to (lord no), but
> > rather the sloppiness of the algorithms used to swap it.
>
> Ever tried a WM device?  Our 624Mhz WM handheld is still painfully
> slow compared to my Centro.  While I think NVFS could have been much
> better designed in terms of stability and availability of diagnostic
> tools (like, how about a way manually shuffle the DBCache or examine
> its contents!), I don't see much of a slowdown except when launching
> Blazer, and I'm not sure that is entirely NVFS-related.

Oh, I know how slow Win-Mob devices are ... I had a foray into "the
other camp" a few years back.  I could not get back to a Palm device
fast enough!  Been watching ever since to see how long it would take
them to get up to a decent speed.  Still waiting.  Not sure I care at
all anymore though.  I've got an N810, and the new Garnett VM beta! :)

And just so you know, that wasn't a complaint against Palm for speed
(which may be the reason they weren't more stringent with the
swapping), but rather for their not coming up with a way to check
those callback addresses and load the bloody page(s)!!  ("chunks".
whatever!)  It just could not possibly have been all that much of a
speed hit, since they know exactly where they are in NVFS!  And even
if I'm wrong about the speed hit, that is what user controlled app
locking is really for, no?

Ken

But, thanks again for the crystal clear explanation!

Any plans to port Teal apps to iPhone?
alan

--- In tealtalk@yahoogroups.com, "Alan" <shrink@...> wrote:
>
> Any plans to port Teal apps to iPhone?
> alan
>
And iTouch??

My Treo randomly reset itself a couple of days ago, and following that
TealBackup (1.70) didn't do its automatic backup. Is this plausible?
Do soft resets clear timers?

Is there any way to automatically reactivate TealBackup after an
upgrade or reset? Might it be possible to check/start it from another
application likely to be used frequently, e.g. TealPhone, or maybe a
DateBk6 plug-in?

Thanks.

TealBackup automatically reschedules itself after a soft reset as part
of its normal reset-handling.

If it isn't rescheduled, here are a few possibilities:

1) The reset may have been a "warm" reset (safe boot) which may have
occurred because the "up" control was held down (by a carrying case
perhaps) when the device crashed/reset.  If so, reset again manually to
restore all your background apps and Tealbackup.

2) If you are using UnCache, make sure TealBackup isn't excluded from
running.  UnCache is a misleadingly named app, and actually just blocks
apps from launching after a reset.

3) Make sure TealBackup's settings are all still there.  If they're
missing, you may have the "Saved Preferences" problem that plagues some
versions of PalmOS, where some preferences go missing after a reset.
You may be able to use "PrefDoctor" or a similar app to prevent this
from happening, or might be able to fix it by copying the Saved
Preferences database to a card with TealMover and then back into memory
again.

-tex
--- In tealtalk@yahoogroups.com, j13d13s <no_reply@...> wrote:
>
> My Treo randomly reset itself a couple of days ago, and following that
> TealBackup (1.70) didn't do its automatic backup. Is this plausible?
> Do soft resets clear timers?
>
> Is there any way to automatically reactivate TealBackup after an
> upgrade or reset? Might it be possible to check/start it from another
> application likely to be used frequently, e.g. TealPhone, or maybe a
> DateBk6 plug-in?
>
> Thanks.
>

Not sure yet.  After we're done rewriting TealAuto and porting it and
some of our other apps to Windows Mobile, we'll revisit the iPhone/iPod
Touch.

The main issue is that we'd like to reuse our existing "C" code and
compile it on the same Windows PCs where our other versions are made,
so they'll all share the same source code, tools, and build system.

My understanding is that iPhone apps are written in Objective-C, a
superset of "C" that came from NeXT that almost nobody uses except
Apple, and that the official SDK requires a Mac.  There appears to be
at least one third-party tool chain that runs under Windows, however,
we just need to find time to try it out.  If successful, we'll try to
port over our TICL library and the apps.

-tex

--- In tealtalk@yahoogroups.com, "mckelveyw" <mckelveyw@...> wrote:
>
> --- In tealtalk@yahoogroups.com, "Alan" <shrink@> wrote:
> >
> > Any plans to port Teal apps to iPhone?
> > alan
> >
> And iTouch??
>

Tex, I have had excellent results with Reset Doctor.

Ben

Another beautiful day in paradise.
-----Original Message-----
From: tealtex <no_reply@yahoogroups.com>
Date: Tuesday, Jun 17, 2008 11:34 am
Subject: [tealtalk] Re: Again, TealBackup doesn't automatically back up
To: tealtalk@...: tealtalk@yahoogroups.com



TealBackup automatically reschedules itself after a soft reset as part
  of its normal reset-handling.

  If it isn't rescheduled, here are a few possibilities:

1) The reset may have been a 'warm' reset (safe boot) which may have
  occurred because the 'up' control was held down (by a carrying case
  perhaps) when the device crashed/reset.  If so, reset again manually to
  restore all your background apps and Tealbackup.

2) If you are using UnCache, make sure TealBackup isn't excluded from
  running.  UnCache is a misleadingly named app, and actually just blocks
  apps from launching after a reset.

3) Make sure TealBackup's settings are all still there.  If they're
  missing, you may have the 'Saved Preferences' problem that plagues some
  versions of PalmOS, where some preferences go missing after a reset.
  You may be able to use 'PrefDoctor' or a similar app to prevent this
  from happening, or might be able to fix it by copying the Saved
  Preferences database to a card with TealMover and then back into memory
  again.

-tex
--- In tealtalk@yahoogroups.com, j13d13s <no_reply@...> wrote:
>
> My Treo randomly reset itself a couple of days ago, and following that
> TealBackup (1.70) didn't do its automatic backup. Is this plausible?
> Do soft resets clear timers?
>
> Is there any way to automatically reactivate TealBackup after an
> upgrade or reset? Might it be possible to check/start it from another
> application likely to be used frequently, e.g. TealPhone, or maybe a
> DateBk6 plug-in?
>
> Thanks.
>

---
Thanks Tex, I'll choose to be cautiously optimistic.
alan


In tealtalk@yahoogroups.com, tealtex <no_reply@...> wrote:
  If successful, we'll try to
> port over our TICL library and the apps.
>
> -tex
>
> --- In tealtalk@yahoogroups.com, "mckelveyw" <mckelveyw@> wrote:
> >
> > --- In tealtalk@yahoogroups.com, "Alan" <shrink@> wrote:
> > >
> > > Any plans to port Teal apps to iPhone?
> > > alan
> > >
> > And iTouch??
> >
>

> 3) Make sure TealBackup's settings are all still there.  If they're
> missing, you may have the "Saved Preferences" problem that plagues some
> versions of PalmOS, where some preferences go missing after a reset.
> You may be able to use "PrefDoctor" or a similar app to prevent this
> from happening, or might be able to fix it by copying the Saved
> Preferences database to a card with TealMover and then back into memory
> again.
>
> -tex

Does PrefDoctor do this any differently (better or worse) than TealScan?

Thanks as always for the very prompt and thoughtful response!
* It wasn't a safe boot - I launched the phone application
   and ding! the phone rebooted. No buttons were pressed.
   I must be using some other badly-behaved application...
* I don't use UnCache or any other hacks
* All the TealBackup settings were still there.

Again, I wonder whether it might be possible to automatically verify
that TealBackup is active from another application likely to be used
frequently, e.g. TealPhone.  I want to completely forget about
TealBackup until that moment when I really need it -- but it seems
that I can't. I hate the idea, but I will probably put a floating
entry in my calendar to remind me to check on it weekly.

Good idea.

I think the proper thing to do would be to give TealBackup an option to
add a message to the Attention Manager whenever a backup is successful,
or if the timer goes off but TealBackup cannot launch itself.  (If
something prevents the timer from going off then of course TealBackup
can't add any messages because none of its code is running).

I'll add this idea to our suggested features list.

-tex


--- In tealtalk@yahoogroups.com, "Jon" <jseder@...> wrote:
>
> Thanks as always for the very prompt and thoughtful response!
> * It wasn't a safe boot - I launched the phone application
>   and ding! the phone rebooted. No buttons were pressed.
>   I must be using some other badly-behaved application...
> * I don't use UnCache or any other hacks
> * All the TealBackup settings were still there.
>
> Again, I wonder whether it might be possible to automatically verify
> that TealBackup is active from another application likely to be used
> frequently, e.g. TealPhone.  I want to completely forget about
> TealBackup until that moment when I really need it -- but it seems
> that I can't. I hate the idea, but I will probably put a floating
> entry in my calendar to remind me to check on it weekly.
>

I notice a big difference between TealLock 7.01 and the previously
released official version. Certain background processes that are set as
AllowedApps don't successfully execute on a consistent basis. The most
obvious one is BackupMan. In that past, it executed successfully at a
particular time each day, locked or unlocked. Now, even if I have
TealLock set to briefly unlock for a period that starts one minute
before BackupMan is set to execute, and end one minute after that time,
    BackupMan only rarely executes successfully at the scheduled time.
With the prior official release of TealLock, BackupMan executed
successfully every time without this setting in place.

>    Posted by: "Ralph Alvy" ralph@... RalphAlvy
>    Date: Sat Jun 28, 2008 9:34 am ((PDT))
>
> I notice a big difference between TealLock 7.01 and the previously
> released official version. Certain background processes that are set
> as
> AllowedApps don't successfully execute on a consistent basis.
>

I just upgraded to that last TealLock & TealBackup. My backups run on
Sundays in the wee hours. This week, my phone was plugged in and I
noticed unusual behavior that I dismissed until I read your note.

My phone was on and there was a birthday message (hidden as a private
appt) and an IM message on the screen. There was also a message saying
something like "About to go back to sleep".

I never saw the latter message before, but....

I don't usually leave my phone plugged in to charge at night and I
have a utility called STOIC (Stay On in Cradle). So, this alone may
have been the cause of the unusual condition in my case.

Paul

Following up on this ... I extended the time from 3 minutes to 10 minutes,
where TealLock will unlock for a brief time if woken up during that time, and
BackupMan seems to be working okay when the device is locked now. Before
7.01, I never had to use this feature for BackupMan.

On Saturday 28 June 2008 09:34:37 am Ralph Alvy wrote:
> I notice a big difference between TealLock 7.01 and the previously
> released official version. Certain background processes that are set as
> AllowedApps don't successfully execute on a consistent basis. The most
> obvious one is BackupMan. In that past, it executed successfully at a
> particular time each day, locked or unlocked. Now, even if I have
> TealLock set to briefly unlock for a period that starts one minute
> before BackupMan is set to execute, and end one minute after that time,
>    BackupMan only rarely executes successfully at the scheduled time.
> With the prior official release of TealLock, BackupMan executed
> successfully every time without this setting in place.