[bookport] Re: Thoughts on BEta 28
- From: "LARRY SKUTCHAN" <lskutchan@xxxxxxx>
- To: <bookport@xxxxxxxxxxxxx>
- Date: Mon, 13 Mar 2006 15:24:36 -0500
Brian:
We really do have the lagging pointer problem fixed. The problem is
that DoubleTalk misses index markers under certain conditions. Those
conditions are that you are in some punctuation and we send certain
punctuation characters with white space around them. You can verify
this by turning on punctuation. You will never see a lagging pointer.
Now, it is possible that we may have missed a punctuation character that
exhibits this problem, but to my knowledge, we have them all covered.
>>> buhrow@xxxxxxxxxxxxxxxxxxxxx Monday, March 13, 2006 3:02:21 PM >>>
Hello. Although I realize the release candidate in the form of
Beta
30 is almost ready to be released, I thought I'd share my thoughts on
Beta
28, because that's what I've been running for the past three weeks, and
I
believe my comments are still applicable to Beta 30, and, thus, the
release
candidate.
Overall, this is a great improvement over the 2.1.0 release
code.
Specifically, the race condition I've complained about on this list
and
privately to Larry is greatly reduced, although it's not gone
completely as
some on this list can attest, and the filesystem handling on large
directories is quite snappy. The spoken minutes timer, which I've
long
used as a "mean time between crashes" feature, reports no crashes in
over
1000 minutes, which is pretty good, since I've made a concerted effort
to
try and crash it with this new beta.
While I do think this code is good enough to release to the
general
bookport population, I still have a couple of issues, which I'll
discuss
below.
1. The time stands still bug.
When playing an indexed audio file, if you press 1 until
Bookport
beeps and positions the file pointer at the beginning of the file, if
you
press 1 again, you get the error tone. However, if you press any
index
motion key which would move you back beyond the beginning of the file
after
you get the error tone, Bookport will, rather than beeping another
error
tone, begin playing the audio file. It will play the audio file all
the
way through until the end of the file is reached or until another key
is
pressed. Now that some work has been done to correct the problem of
the
spoken minutes timer incorrectly reporting too much elapsed time, it's
worth noting that during all of this play, the spoken minutes timer is
not
incremented at all, regardless of how long the Bookport plays the
audio
file. It's also worth noting that, although I realize APH feels
they've
resolved the trailing pointer problem by jiggering the way indexes are
made
by the transfer utility, I'm fairly sure that this bug, and the
trailing
pointer bug are closely related some how. I base this on the fact
that
when one of my text files begins to demonstrate the trailing pointer
bug,
if I switch to an audio file, jump to the beginning of the file, and
then
induce the Bookport to begin playing the file in this non-orthodox
manner,
then when I switch back to the text file, the trailing pointer problem
ceases.
2. The trailing text pointer problem.
Again, although I realize APH feels they've addressed this issue
by
changing the way the transfer utility indexes text files, I'd really
like
to encourage APH to see if they can't find the root of the problem,
rather
than fixing the symptom. The reason is that I believe the trailing
pointer
bug is a symptom of some more fundamental bug in the firmware code,
which
may lead to manifestations of more serious bugs later on down the road.
I
think some register is getting corrupted somewhere in the code, and
I'm
concerned that this register, or memory location, or what ever it is,
might
be used for something more important later, and thus cause bigger
problems
down the road. In other words, it has the feel of random bits
flipping
inside my Bookport, and that make sme nervous.
Again, I want to say that this beta represents a vast
improvement over
the release code, and I do think it's ready for prime time use, with
the
minor caveats above.
-Brian
Other related posts:
