Quote from: timwessman on September 21, 2013, 05:45:31 pm

Quote

Also is the calc really sold with OS 0.025.5106? A version number under 1.0.0 often means it's a beta...

Nope. That is a common convention, but really has nothing to do with the state of things. That whole menu is for factory use and they control numbers and versions there. I honestly have no idea what they really are using there or what it is supposed to mean. I suspect it has something to do with bootloader stuff.

Quote from: iconmaster on September 21, 2013, 12:19:48 pm

some features feel incomplete

Examples please?

I'd like to add onto the incompleteness.

This certainly feels like a beta. There are certain things from a UX standpoint that don't seem to work as expected. I've screen-capped a couple of them using the connectivity software:

After trying to type the following integral in, I got the weirdest, most undescriptive error that I've ever seen on a calculator. Trying to copy/paste yielded the same result, yet retyping the function gave the correct result.


Trying to find the integral of any trig function (as far as I can tell, only tested it with sin/cos) yields the following message on the screen. I understand that the calculator uses taylor series to approximate an integral, but that message doesn't give me any useful information as a user, and since it requires user interaction to dismiss (it doesn't go away automatically), it is very annoying.


In the below picture, both the function on the top and the bottom are equivalent, but the Prime refuses to evaluate the top symbolically.


Implicit multiplication seems to work randomly. 13x seems to work, but 13sin(x) throws yet another, even less descriptive error.


Other implicite multiplications that seem to not work as expected are π(10) (This returns a bad argument value error, which seems to suggest that it's intended for the Prime Counting Function, but that doesn't work as expected either.). Various other implicit multiplications either error, or do not work as expected. In the same picture, prettyprint (textbook notation?) doesn't seem to work for e. instead of typing something like e^x (or even the nicer e^x), it does exp(x)



I love the prime, and the graphing is very fast, but these are glaring errors that really take me out of the entire experience. Also, if I wanted to enclose a part of an expression in parenthesis, i have to follow these steps:

Navigate to beginning of part I want enclosed
Push () button
Push right dpad
Push backspace
Navigate to end of part I want enclosed
Push () button
Push backspace

With the Nspire, these were the steps:
Highlight part I want enclosed
Push ( or ) button

OR

Navigate to beginning of part I want enclosed
Push ( button
Navigate to end
Push ) button

Finally, if I make an error in my input (forget the dx for the integral, forget to square something properly, etc), the calculator doesn't allow me to fix it, it reformats the function in some random way, and then throws up an error.

I'm trying really hard not to compare this to my Nspire from a UX point of view, since I recognize the need to differentiate from one's competitors, and a direct copy would be boring, but when these UX decisions affect the usability of the calculator, it gets annoying.

Quote from: timwessman on September 21, 2013, 05:45:31 pm

Nope. That is a common convention, but really has nothing to do with the state of things...

Oh, that's good to know! Look at me here, always assuming the worst.

Quote from: timwessman on September 21, 2013, 05:45:31 pm

Examples please?

Eh, it's just a sort of an impression. Really, I'd be writing more about suggestions I have than things that are really missing. But, you did ask, so here goes:
* The app structure doesn't really have an HP calculator feel. Maybe if there was app 'folders' and/or an ability to hide apps?
* The "online" Help has a lot of typos, not to mention the accidental HP38gII syntax.
* Like the first point, there's no good way to organise your variables.
* There's no calculator-to-calculator linking. Even if the physical manual tries to claim there is ("Documentation which mentions sending data directly from one HP Prime to another should be disregarded" the sheet of paper that came with my Quick Start Guide says).
* All the undocumented CAS shortcut functions, described here. Some of them can cause crashes!
* Some basic apps we'd like are not here, like 3D graphing or periodic tables. No extra utility apps either.

All in all, it's a fine release, sure. But it just feels that much more limited than the HP calculators I'm used to (I mean heck, my 50g came with a built-in ARM assembler!). My hypothesis is that HP is just playing nice in the software until they get College Board approval, at which point I hope they'll add in more freedom to the OS. Of course, the College Board won't be easily convinced, so this could take a while to happen.

While the fire for random suggestions are hot, here's another one. I know how you've been dealing with text highlighting since the beginning of time: Using the Mark system. Myself, I'd think one could implement a text-highlighting system closer to the TI-89's, using the Alpha key or similar... It'd make text highlighting a lot more natural.
Also, there's no way to select the item on drop-down input boxes without touching the screen. I'd think you could hit ENTER to open up the dropdown, but no.

Here's a good example of the oddities of the CAS. Let's try using the @ sign in an expression. Nothing in the documentation says anything about the @ sign in an expression, so it'll be a syntax error if we use it, right?





...Wrong. It instead gives us double_deux_points, which is something I've never heard of, and the Catalog hasn't, either.



Copying this mysterious input does not help resolve questions. It's just another undocumented CAS convention. Re-inputting the same exact output gives us just 5, but copying the 'just' 5 gives us at::5 again. Will the wonders never cease?

EDIT: A deux point is French for a colon? Well, that's nice and literal.

EDIT 2: I installed the SDK OS. Well, one difference is this:


EDIT 3: I cannot seem to reproduce the above behavior, so it might not be an SDK thing. Just a bug thing. Also, it seems that ON+Help seems to do something, but I can't tell what. Anyone else with a non-SDK Os, does ON+Help not give you the Help screen like it should if ON+Help didn't do anything?

EDIT 4: Okay, on another note, I extracted all the known localized strings using the STRINGFROMID function. I thought there might be some secrets or whatnot in there. It's available here. Some entries in this list implies that the HP Connectivity Kit program exists inside the calculator, oddly enough.

Quote from: timwessman on September 21, 2013, 05:45:31 pm

Quote from: iconmaster on September 21, 2013, 12:19:48 pm

some features feel incomplete

Examples please?

For exemple, I've found a limitation with the sequence application.
By default it starts sequences with rank 1, but it doesn't let you set the sarting rank to 0.


This is a big problem for France, as the sequences are a very important topic in high school, and we define them as starting at rank 0 - and so start many sequences in problems.

Other calculator manufacturers (TI & Casio) set the default starting rank to 1, but let the user enter 0.



And now what seems to be a bug of the sequence application - I don't know if it's called like that in english, but it seems you can't define cross-sequences:


U1 sequence is disabled and unusable.
I suppose the app checks and sets the sequences sequentially - so U1 is checked first but rejected as it depends upon U2 which is not defined at that time.


The same problems seem to be present with the HP-39gII sequence app. (I don't have the calculator - I could only test with the emulator)


I've reported both problems to HP-France, but I don't know if they did translate and forward them.

Bug 1:

Code: [Select]

WHILE 1 DO
FOR X FROM 0 TO 320 DO
FOR Y FROM 0 TO 240 DO
GETPIX_P(X,Y)▶P;
RECT_P(X-INT(.05*Y),Y-INT(.05*X),X-INT(.05*Y),Y-INT(.05*X),P,P);
END;
END;
END;After running this program once and letting it loop through the While at least once, the next time you run it you will get an argument error until rebooting the emulator. In certain cases, the argument error will occur even if you didn't let the While loop go on at least once. Also, this causes the Tunnel game to stop working.


Bug 2:

Code: [Select]

DIMGROB_P(G1,120,48,RGB(255,0,0,128));
TEXTOUT_P("HELLO",G1,0,0,7);
FOR Z FROM 1 TO 800 DO
RECT_P(RGB(Z/4,Z/4,255));
BLIT_P(G0,0,0,Z,Z,G1,0,0,10,8,RGB(255,255,255,128));
WAIT(0.03);
END;
If you run this program then quit with ON, on the screen saying the error user aborted message, if you click the LCD, then press a key and land on a different screen, if you decide to go back to the program list screen, pressing the arrows will no longer move the selection around (although it will still let you choose a different program. you just won't see what you have selected).

Here's a bug: When the calculator is in the Dark GUI setting, the labels created in the Numeric view of the Geometry app appear behind the Light GUI's background.



Another one: In the emulator, the ? prefix in the CAS is a synonym for xcas's findhelp function, which does nothing on the calc. In the physical calculator, however, using it causes a crash.

I know that this has been said over and over again, but I'll say it again here:

The screen's got multi-touch. Why don't we have pinch to zoom on the graph? Or we could do taps too, two fingers for zoom in, 3 fingers for zoom out. it's just intuitive to have a touchscreen and to be able to pinch to zoom.

Another bug. This time, it's with units. I think the image says it all:




And another suggestion, this time with the Views key. Okay, so we have Numeric, Plot and Symbolic views, right? Well, we have two different uses of the term view: One, in the 3 basic screens, and another in how these screens can be looked at. This is needlessly confusing terminology. Maybe if the Views key lived up to its name by displaying all the available views, including the views you can get with the current Views key. So, for example, the Function app Views menu would look like this:

• Symbolic View
• Plot View: Autoscale
• Plot View: Decimal
• Plot View: Integer
• Plot View: Trig
• Numeric View
• Split View: Plot/Plot
• Split View: Plot/Numeric

Or maybe those 'plot view' and 'split view' menus could be sub-menus. I don't know, you're the HP official here Tim.

This one's more a suggestion than a bug: Despite being in CAS mode, using the Define box only creates 'normal' functions, not xcas ones.

Another bug: This program errors with "Invalid input" when I hit ENTER, yet works fine in debug mode:

Code: [Select]

EXPORT CURSOR()
BEGIN
LOCAL X,Y,K,C,P;
0▶X;0▶Y;0▶P;
WHILE 1 DO
 //WAIT()▶K;
 WAIT(.05);
 IF ISKEYDOWN(4) THEN
  RETURN {X,Y};
 END;
 IF ISKEYDOWN(2) THEN
  GETPIX_P(G0,X,Y)▶C;
  BITNOT(C)▶C;
  PIXON_P(G0,X,Y,C);
  Y-1▶Y;
  GETPIX_P(G0,X,Y)▶C;
  BITNOT(C)▶C;
  PIXON_P(G0,X,Y,C);
 END;
 IF ISKEYDOWN(12) THEN
  GETPIX_P(G0,X,Y)▶C;
  BITNOT(C)▶C;
  PIXON_P(G0,X,Y,C);
  Y+1▶Y;
  GETPIX_P(G0,X,Y)▶C;
  BITNOT(C)▶C;
  PIXON_P(G0,X,Y,C);
 END;
 IF ISKEYDOWN(7) THEN
  GETPIX_P(G0,X,Y)▶C;
  BITNOT(C)▶C;
  PIXON_P(G0,X,Y,C);
  X-1▶X;
  GETPIX_P(G0,X,Y)▶C;
  BITNOT(C)▶C;
  PIXON_P(G0,X,Y,C);
 END;
 IF ISKEYDOWN(8) THEN
  GETPIX_P(G0,X,Y)▶C;
  BITNOT(C)▶C;
  PIXON_P(G0,X,Y,C);
  X+1▶X;
  GETPIX_P(G0,X,Y)▶C;
  BITNOT(C)▶C;
  PIXON_P(G0,X,Y,C);
 END;
 IF ISKEYDOWN(30) THEN
  REPEAT
   MOUSE()▶P;
   P[1]▶P;
  UNTIL P[5]==#1;
  B→R(P[1])▶X;
  B→R(P[2])▶Y;
 END;
 END;
END;
END;


Today's bug: Okay, this one's really hard to describe, as it's apparently non-deterministic. It's similar to DJ_O's issues. At some point in time, SUBGROB and DIMGROB just started failing to work. They were working as documented earlier; now they just give me an 'invalid input' error whenever I try to use them. In one case, the use of SUBGROB, instead of giving an error, just stopped program execution dead and returned immediately, inside an infinite loop no less! Debugging them is even worse. Sometimes when going over one of the error statements in the debugger, the calculator will just crash, but not all the time.

Today's suggestion: It's about debugging. When you hit 'Cont' on the debugger, it just goes along like normal. If there's an error, it gives the message as usual, but the error's almost useless as no line of code is pointed to. Getting the line of errors is the point of debugging! Maybe if Cont just stepped through really quickly instead of just running the program as usual, we could get the line of the errors without pressing 'Step' a lot or setting up a lot of breakpoints.

Not really a bug, but did you ever have plans to add a mode in the emulator that makes its speed identical (with no exception) to the real calculator (kinda like WabbitEmu for TI calcs)? The issue is that if people use the emulator to develop programs that are dependent on a specific speed, we have to send them on the real calc over and over because the emulator speed is misleading.

I had a bug for today, but I forgot it in the hours I've been away from these forums. So I'll just tell you some places where the Help doesn't show up!

Today's bug: There's no Help section when pressing the Help button for the Characters menu, any of the coose boxes (templates, the ! menu, etc.) or the Memory Manager.

Today's Suggestion: There's only 10 graphic variables ever?? If we want to use sprites, we can only use the 10; if we want more, we have to hot-swap them out using DIMGROB, which can't possibly be memory/speed efficient.