▼
Posts: 247
Threads: 26
Joined: Oct 2007
My 15c LE arrived today. (Whee!) Part of putting it through the initial paces has involved converting my typical binomial probability distribution program. It's nothing fancy - it just takes a few input parameters, and returns a probability:
t: Number of trials
z: Probability of individual success
y: Min number of successes desired
x: Max number of successes desired
If I run it with inputs 3, 1/3, 1, 3, I get the correct result, but the display flashes like there was an overflow somewhere. If I step through the program, using the same inputs, this doesn't happen, and I still get the same correct result..
I'm not quick to shout "BUG!" so what dumb thing am I likely doing wrong?
▼
Posts: 875
Threads: 37
Joined: Jul 2005
Could we see your program listing?
▼
Posts: 247
Threads: 26
Joined: Oct 2007
Here's a hasty transcription. The program still needs some optimization, obviously.
LBL B
STO 4
RDown
STO 3
STO 0
RDown
STO 2
RDown
STO 1
CLx
ENTER
ENTER
ENTER
LBL 7
RDown
RDown
RCL 1
RCL 0
Cy,x
RCL 2
RCL 0
y^x
*
1
RCL- 2
RCL 1
RCL- 0
y^x
*
+
1
RCL+ 0
STO 0
RCL 4
x>=y
GTO 7
CLx
RDown
CLx
RDown
RTN
▼
Posts: 125
Threads: 5
Joined: Jun 2008
Is 0.7037 the correct result?
▼
Posts: 247
Threads: 26
Joined: Oct 2007
Yes, I'm pretty sure that's correct. I get that result whether I run the program, or step through it, but the display only flashes if you run it normally.
And this is even weirder. If you run it with these inputs, the display doesn't flash: 3, 1/3, 0, 3
Try as I might, I can't find any overflow happening in the program.
▼
Posts: 125
Threads: 5
Joined: Jun 2008
I used my classic HP-15C to run your program. No flashing display!
This may be another known bug.
Anyway, I got 1 from your input.
Edited: 16 Sept 2011, 7:46 a.m.
▼
Posts: 247
Threads: 26
Joined: Oct 2007
Just to clarify my tests a bit...
3, 1/3, 1, 3: Result: .703, flashing display
3, 1/3, 0, 3: Result: 1, no flashing display
The flashing display "error" will only happen if I run the program with "f B" (or just "B" in USER mode), or "GSB B". If I run it by pressing "GTO B R/S", the display does NOT flash after executing. Nor does the display flash if I step through the entire program with SST.
So, am I screwing something up, or is it a calculator bug? I tried this with the included 15c emulator for Windows, and the results appear the same.
Posts: 875
Threads: 37
Joined: Jul 2005
I also confirm the behavior, and also do not get it on my 15C Classic. I note that if I set the program pointer to Label B with GTO B, then press R/S, the flashing display does not occur (with the 3, 1/3, 1, 3 inputs). I do not believe there is any overflow in your program. For a true overflow, say tan(90), you get a 9.999999 e99 flashing display. Pressing the back-arrow key stops the flashing but does not clear the display. With your program, pressing the back-arrow key stops the flashing and clears the display. Perhaps this is related to the flashing display behavior reported by Katie a day or two ago, i.e., if that problem is found and fixed maybe this one will also go away.
Edited: 16 Sept 2011, 8:27 a.m.
▼
Posts: 247
Threads: 26
Joined: Oct 2007
Here's a much simpler test case that exhibits the same behavior.
LBL A
STO 0
LBL 9
8
ENTER
3
Cy,x
DSE 0
GTO 9
RTN
If you put 3 on the stack and run it with "f A" or "GSB A", the screen will flash. Other integer inputs don't seem to cause this. Something about running Cy,x or Py,x exactly 3 times in a loop that's started with GSB seems to cause it.
▼
Posts: 1,477
Threads: 71
Joined: Jan 2005
This is definitely a new bug you've found.
I think we need to start a bug report list for the 15C LE and put it somewhere that's accessible. Are these all of them so far:
1) PSE run more than once in a program blanks display and run even once causes display flashing issues.
2) When Cx,y/Px,y are run in a program multiple times with certain values (e.g., 8 ,3) it causes display flashing issues.
3) Executing the self tests documented in the manual can corrupt program memory.
-Katie
▼
Posts: 850
Threads: 10
Joined: Mar 2009
Also the current draw you found:
4) High current draw during key press/hold
-B
▼
Posts: 1,477
Threads: 71
Joined: Jan 2005
Thanks.
I'll put this in an article and edit it as needed.
-Katie
Posts: 247
Threads: 26
Joined: Oct 2007
Fortunately, the known bugs mostly seem to be minor cosmetic issues so far, with the exception of the self-test hosing memory (and with only a few hundred bytes of RAM, even the impact of that one is limited).
The PSE bug is annoying, but I personally don't use that instruction much when programming. I prefer to have a flag switch the program between "run as fast as possible until completion", and "stop and show intermediate results until R/S is pressed to continue".
So, how can I get a cable to flash the firmware when an update is inevitably released? :)
(Just found another one: Do 90!, ENTER. The display will keep flashing even if you start entering other numbers.)
Edited: 16 Sept 2011, 11:07 a.m.
▼
Posts: 1,477
Threads: 71
Joined: Jan 2005
Quote:
(Just found another one: Do 90!, ENTER. The display will keep flashing even if you start entering other numbers.)
This is normal.
▼
Posts: 247
Threads: 26
Joined: Oct 2007
Okay. Then not a bug - just odd. :)
Posts: 127
Threads: 21
Joined: Jul 2006
I believe Katie is right!
HP should be advised about the bugs on HP 15C AE.
I ordered one from Samson Cables, but, sincerely, I almost removed my order yesterday, when I saw the bug concerning Cx,y and factorial functions and even the printing of these functions on calculator's face plate. I only did not remove my order because my name was also in that list of buyers for Bring HP15C Back movement.
I live in Brazil and if I have to send the calculator back to HP, it will be very expensive and a dangerous operation.
After all, I believe that a calculator on which you can't trust the results doesn't worth one little penny - I apologize the effort spent on it, but it is uselless. Very hard and heavy words, sorry...
I was very inclined to buy one 30B and rewrite it with WP34S and wait or help (as I can) on developing matrices routines for it.
Edited: 16 Sept 2011, 11:47 a.m.
▼
Posts: 247
Threads: 26
Joined: Oct 2007
Quote:
HP should be advised about the bugs on HP 15C AE.
Yes, then we can get a firmware update.
Quote:
After all, I believe that a calculator on which you can't trust the results doesn't worth one little penny - I apologize the effort spent on it, but it is uselless. Very hard and heavy words, sorry...
Not by a long shot. The known bugs so far are largely just goofy cosmetic issues. So far I don't believe anybody has found a way to make the 15c LE produce wrong results. Cy,x and Py,x appear to work perfectly fine - it's just that for some reason, calling them three times within a loop can make the display flash as though there had been an overflow error (even though the results are still correct).
The memory loss from running the self test should be addressed, but it's not nearly as bad as it sounds. For one, you only stand to lose a few hundred bytes of memory, and secondly, what are the chances of running the self test accidentally? Compare this to the freeze/memory loss problem in the 35s, which could rear its head as the result of a poorly coded loop. And in that case, you stand to lose up to 32 KB of programs and data. Not fun! (I speak from experience there.)
▼
Posts: 127
Threads: 21
Joined: Jul 2006
Thanks, Dave!
I really hope not be disappointed with this so long time waited reborn ...
Artur
Posts: 325
Threads: 18
Joined: Jul 2006
Just to complicate the issue, I don't get a flashing display no matter how I run this program on my HHC2010 15C+.
I'm glad I hadn't gotten around yet to reflashing my 15C+ with the ROM image Gene sent me. (Thanx, Gene. I'm still holding on to that file.)
▼
Posts: 1,477
Threads: 71
Joined: Jan 2005
Quote:
Just to complicate the issue, I don't get a flashing display no matter how I run this program on my HHC2010 15C+.
I'm not surprised and there may be other beta-test versions around as well. That's why I put the firmware date at the top of bug list article, it's very important to cite the firmware version when claiming a bug. (I just hope that there is never more than one version released per day.)
-Katie
Edited: 16 Sept 2011, 2:45 p.m.
|
