[BUG] Spreadsheet Copy Multiple Cells Causes Crash

[Michael de Estrada]

I actually reported this bug previously in the old forum, but it received few responses and I just want to be sure that HP has seen it for a fix in the next firmware release. This bug only occurs in the physical Prime and not the Emulator.

1. Save the Spreadsheet App as a new file, say Test. Entry mode is either algebraic or Textbook.
2. Select blank cells E1 through M1. Press Shift Copy. The Copy confirmation is shown. Nothing bad happens.
3. Enter formulas and values into cells E1 through M1: =C1*D1,=E1*D1^2/12,=E1*C1^2/12,=E1*A1,=E1*A1^2,=E1*B1,=E1*B1^2,=E1*A1*B1,0. Zeros appear in all the cells as expected, since cells A1 through D1 are blank.
4. Select cells E1 through M1 (9 cells). Press Shift Copy. The calculator reboots and resets !
5. Select cells E1 through I1 (5 cells). Press Shift Copy. The copy confirmation is shown. Select cells below and perform Shift Paste. All is good and the contents of the cells pasted are correct.
6. Power Off to save the work and On again.
7. Select the remaining cells J1 through M1 (4 cells). Press Shift Copy. The calculator reboots and resets !
8. Select cells J1 through K1 (2 cells). Press Shift Copy. This works and the pasted cells are correct.
9. Power Off to save the work and On again.
10. Select the remaining cells L1 through M1 (2 cells). Press Shift Copy. The calculator reboots and resets !
11. Select cell L1 (1 cell). Copy / Paste works.
12. Select remaining cell M1 (1 cell). Copy / Paste works.

[Han]

M1 looks suspicious given that there is also a matrix of the same name.

[Michael de Estrada]

M1 is just the cell address. It is not a variable.

[Han]

(12-18-2013 10:45 PM)Michael de Estrada Wrote:  M1 is just the cell address. It is not a variable.

While you are correct, my response was a guess as to why the crash was happening -- that is, there may be internal conflicts with M1 the cell vs. M1 the built-in global matrix variable.

[Michael de Estrada]

If this were the case, why does this problem exhibit itself only in the physical Prime and not the Emulator ? Also, note that in the last step 12, selecting cell M1 and then <shift> Copy does not crash the calculator and the operation does work.

Edit. Let me take back some of what I just said. You just might be onto something here. I tried selecting cells M1 thru M9 along with the corresponding cells L1 through L9, and pressing Copy crashed the calculator in every case. However, selecting cells L10:M10 does not crash the calculator. All these cells contain the same formulas, but the difference is that there is no built-in global variable M10, whereas M1 through M9 are global matrices. I still can't understand why this problem only affects the physical Prime though.

[Han]

(12-19-2013 01:19 AM)Michael de Estrada Wrote:  If this were the case, why does this problem exhibit itself only in the physical Prime and not the Emulator ?

The two are not identical. That one behaves differently from the other should come as no surprise. The binaries are for different architectures.

Quote:Also, note that in the last step 12, selecting cell M1 and then <shift> Copy does not crash the calculator and the operation does work.

If you notice, though, the column names are lower-cases for g, m, l, and z because M1 -- the cell -- would conflict with the built-in matrix name M1 (similarly for G0-G9 the graphics objects, L0-L9 for lists, and Z0-Z9 for complex numbers). So the hunch is that there is probably some code in the firmware that did not get updated to account for this workaround. But like I said, it was a guess -- nothing more. That you can copy a single cell (M1) is not too strange because it may be using a different internal call to copy a single cell.

There is another quirk:

Enter into any cell =A1 and you don't see a zero, unlike the other cases. However, if you enter =A1*1 then you get 0.

[ArielPalazzesi]

(12-19-2013 01:44 AM)Han Wrote:  There is another quirk:

Enter into any cell =A1 and you don't see a zero, unlike the other cases. However, if you enter =A1*1 then you get 0.

Hi Han!
That seems to be a question related to the format of the cell. When there is a zero (which should be the default for all cells), these appear as empty, not to have the whole sheet full of zeros.
I am not a user of that application, but I think that using some of the formatting options you should be able to avoid such behavior.

A hug.

[Tim Wessman]

Well, the way spreadsheets work(all? or most?) is an undefined item has a "special 0" that is in an "empty" cell. Try the same thing in excel or other spreadsheets.

The question here is are you concerned that it doesn't show anything for =A1? Or that the 0 only appears with =A1*1? Or what is the item of discussion. Feeling a bit confused.

[Han]

(12-19-2013 04:53 PM)Tim Wessman Wrote:  Well, the way spreadsheets work(all? or most?) is an undefined item has a "special 0" that is in an "empty" cell. Try the same thing in excel or other spreadsheets.

The question here is are you concerned that it doesn't show anything for =A1? Or that the 0 only appears with =A1*1? Or what is the item of discussion. Feeling a bit confused.

Aside from the crashes reported by Michael, the =A1 issue is simply that it does not behave in the same way as =A1*1 when cell A1 is undefined. =A1 leaves the contents blank whereas =A1*1 inserts a zero. So the issue is that there is some inconsistency (I personally don't have a preference one way or the other).

[cyrille de brébisson]

Hello

(12-19-2013 01:19 AM)Michael de Estrada Wrote:  If this were the case, why does this problem exhibit itself only in the physical Prime and not the Emulator ?

Because the emulator runs on an Intel CPU which will do unalligned memory read/write of words while an ARM will (unless exception code is written) seg fault on it...

Among the couple of items which differ between intel and ARM are:
- unalligned pointers access -> crash
- shift by a negative value (a>>b when b<0)
- a%b when a and/or b are < 0 which leads to different results..

The first one is the cause of the HW crash. the 2 others are for your general information:-)

Thanks for the bug report. It is beeing taken into concideration.

Cyrille