newRPL: [UPDATED April 27-2017] Firmware for testing available for download

[Claudio L.]

By an overwhelming number of requests (ok, just one but looks better that way) I've made available for download an early firmware image for the adventurous.
It can be downloaded here:
UPDATE: ** USE LINK BELOW FOR LATEST DEMO IMAGE ** - SEE POST #39, #70, #118, #159, #184, #245, #267, #274, #313, #329, #349 #391 #436 and #500 BELOW FOR LATEST CHANGES

newRPL firmware demo

Update: The documentation project is well underway! If you downloaded the ROM above, don't forget to check out the new wiki at:

The newRPL Documentation Project wiki


Update: The latest image is still alpha quality, but it's not as green as it may seem from the old post. It includes many features and can work as a reliable scientific calculator. Much work remains to be done, but this image has improved much since it was originally posted. For more updated information please refer to the project website:

newRPL Project Website

OLD AND OUTDATED INFO HERE.... (original post)
---------------------------------------------------------
[/b]

Everyone is invited to test-drive it and provide feedback. Being an early demo there's many things that don't work, some because they aren't implemented yet, and some because they are buggy. I'd appreciate comments on the latter case so we can start debugging this into a more polished product.

What to expect and not to expect:
* There's no menus, but most commands can be typed in alpha mode.
* The VARs menu is fully functional. It can be hidden/shown by long-pressing VAR.
* No status area yet
* Command line is not multiline yet, and there's no begin/end/cut/copy/paste. Skip word, go to the end, etc. are mostly not implemented yet.
* Keyboard is mostly functional, but there's keys not bound yet to their commands (like HOME or UPDIR, you have to type them).
* The new alpha mode will seem very strange at first, but will quickly grow on you.
* The STO/RCL function was moved to the HIST/CMD location in the keyboard due to the new VARs menu keys interfering with it. Everything else stayed where the 50g had it.
* The OFF function was moved to the battery compartment for now (pull a battery!). The OFF key performs a Warm-Start due to technical difficulties.

Feel free to explore the docs here, and the detailed status to see which commands are already implemented here.
There are a few more commands (several that deal with matrices) that were implemented but not shown in the list yet.

To install the rom, simply put it on an SD card together with a file called 'update.scp' that contains the text 'newrplfw.bin' inside (no quotes). Then do a normal ROM update and it will pick it up. Upon reset, you'll get a confirmation screen that warns you that the initial install will wipe out all RAM. This is your last warning before you lose anything that wasn't backed up.
To restore to stock firmware, just do a paperclip reset and re-flash the stock rom, but you'll have to use the 2MB special version that's in the .zip file to get Port 2 working again.

Let's concentrate the feedback and questions on this thread.
Enjoy! (and report back!)

[Guenter Schink]

Ok,

I've done it. After the first flashing process (shown to be successful), the 50G wouldn't start. So I re-flashed the original firmware, just to make sure, I had the procedures right. The second flashing was successful though. Don't know what happened the first time.

Now it's up and running. Time to play around.

First observation. The "transfer arrow" in the upper right hand corner is flashing all the time. Perhaps something to do with the nasty "OFF" bug.

Numbers with decimals are terminated by another dot.

Quote:* The VARs menu is fully functional. It can be hidden/shown by long-pressing VAR.

I can see the Vars when the vars aren't hidden, but how to access them other then typing their names?

The OFF-function works flawlessly

There will be at least one NR50G at the HHC. *

Günter

* I selected NR50G to express it is NewRpl50G, I hope you like it.

[Claudio L.]

(09-04-2015 09:05 PM)Guenter Schink Wrote:  Ok,

I've done it. After the first flashing process (shown to be successful), the 50G wouldn't start. So I re-flashed the original firmware, just to make sure, I had the procedures right. The second flashing was successful though. Don't know what happened the first time.

Now it's up and running. Time to play around.

First observation. The "transfer arrow" in the upper right hand corner is flashing all the time. Perhaps something to do with the nasty "OFF" bug.

It's a "heartbeat", done on purpose during early development to know if the machine crashed. As long as it blinks, I know it's still alive. Lately it hardly ever crashes so it's not so useful anymore. It will be removed eventually. Same thing with the keys that are not yet defined, it shows a key-code on the status area, that's normal for now.

(09-04-2015 09:05 PM)Guenter Schink Wrote:  Numbers with decimals are terminated by another dot.

Not a bug at all. newRPL adds a trailing dot to approximated numbers. It's explained in the docs, and there's also a thread about it earlier this year.
Basically, if you type a number without a trailing dot, it's assumed to be exact. If you add a trailing dot, it's assumed to be approximated.
For instance, if you type '3/4', then EVAL will remain as a fraction because both numbers are exact. If you type '3./4' then EVAL, it will compute the division and show 0.75., since it already knows the number is approximated. This gives you control of what gets simplified in an expression, just by adding a dot.
If you do the division 3 4 / will return 0.75 (also exact, no trailing dot). But if you do 1 3 / you'll get the trailing dot, since the machine could not compute the result exactly.
Eventually, when exact mode is implemented, this will make more sense.

(09-04-2015 09:05 PM)Guenter Schink Wrote:  

Quote:* The VARs menu is fully functional. It can be hidden/shown by long-pressing VAR.

I can see the Vars when the vars aren't hidden, but how to access them other then typing their names?

I'm not sure I understand what you mean. The 6 keys (APPS, MODE, TOOL, VAR, STO and NXT) act like F1 to F6 but for the lower menu. So you always have access to your vars, no matter which menu you are in (the top menu will change, but your vars will always remain in the lower menu).
Just use those keys to access your vars, they work normally (unshifted to EVAL, shifted to RCL or STO, if you are in the command line they add the name, etc). Even if you make the menu invisible, those 6 keys still give you your variables.

(09-04-2015 09:05 PM)Guenter Schink Wrote:  The OFF-function works flawlessly

There will be at least one NR50G at the HHC. *

Günter

* I selected NR50G to express it is NewRpl50G, I hope you like it.

I do like it. Keep playing with it and report any findings.

[Guenter Schink]

(09-04-2015 09:50 PM)Claudio L. Wrote:  VAR, STO and NXT) act like F1 to F6 but for the lower menu. So you always have access to your vars, no matter which menu you are in (the top menu will change, but your vars will always remain in the lower menu).
Just use those keys to access your vars, they work normally (unshifted to EVAL, shifted to RCL or STO, if you are in the command line they add the name, etc). Even if you make the menu invisible, those 6 keys still give you your variables.

I guess I have to do a lot of reading to review your advanced concepts before placing questions/observations having been answered before. The concept of the vars-keys escaped me.

Quote:

(09-04-2015 09:05 PM)Guenter Schink Wrote:  The OFF-function works flawlessly

There will be at least one NR50G at the HHC. *

Günter

* I selected NR50G to express it is NewRpl50G, I hope you like it.

I do like it. Keep playing with it and report any findings.

Günter

[DavidM]

Any reason this wouldn't work on a 49g+? I'd rather leave my 50g with the 2.15 firmware for now and test out NewRPL on the 49g+ if possible.

[Claudio L.]

(09-04-2015 11:20 PM)DavidM Wrote:  Any reason this wouldn't work on a 49g+? I'd rather leave my 50g with the 2.15 firmware for now and test out NewRPL on the 49g+ if possible.

I think it should work fine (but is not tested).
There are very minor differences in the hardware that could cause some annoyances. The 49G+ had 3 batteries instead of 4, so the low battery warning probably needs different calibration.
Debouncing the keyboard was originally calibrated for hpgcc on a 49G+, so we can reasonably expect good overall keyboard "feel", but it's an area to be tested too.
Try it, worst case you can re-flash the stock rom if it doesn't work, so you have nothing to lose (but time).

[Claudio L.]

(09-04-2015 11:02 PM)Guenter Schink Wrote:  I guess I have to do a lot of reading to review your advanced concepts before placing questions/observations having been answered before. The concept of the vars-keys escaped me.

Günter

Sounds like you are apologizing for asking the questions. You asked about an undocumented new feature that I finished coding less than 2 weeks ago, it's a perfectly valid and pertinent question!
I did a search, and the VARS menu behavior isn't explained anywhere but in some random posts as a concept being discussed over a year ago. From concept to fact a lot happened, and precise information on how it actually works doesn't really exist.
Information is very limited in the documents I wrote, I'm not very proud of those docs but it's what I could do in the very limited spare time I have.

[matthiaspaul]

(09-05-2015 12:48 AM)Claudio L. Wrote:  The 49G+ had 3 batteries instead of 4, so the low battery warning probably needs different calibration.

I don't own a 49g+ so I don't know for sure, but since the ADC won't be able to cope with the full battery voltage anyway, the hardware will most probably contain some kind of voltage divider to bring down the maximum possible voltage to an acceptable voltage for the analog input. If so, the hardware developers might have adjusted the voltage divider so that the different nominal battery voltage would be transparent to the controller.

Quote:Debouncing the keyboard was originally calibrated for hpgcc on a 49G+, so we can reasonably expect good overall keyboard "feel", but it's an area to be tested too.

Don't they both run at 75 MHz by default? If so, keyboard debounce timing should be the same.

AFAIK, the 49g+ uses a Samsung S3C2410X01, the 50g a S3C2410A, but this shouldn't make much difference.

Greetings,

Matthias

[Zweb]

Good news!
I've translated and posted it on cnCalc (an Chinese calculator forum): http://www.cncalc.org/thread-11707-1-1.html , and I will translate the feedback to English and post here(if any).

[Claudio L.]

(09-05-2015 01:52 AM)matthiaspaul Wrote:  I don't own a 49g+ so I don't know for sure, but since the ADC won't be able to cope with the full battery voltage anyway, the hardware will most probably contain some kind of voltage divider to bring down the maximum possible voltage to an acceptable voltage for the analog input. If so, the hardware developers might have adjusted the voltage divider so that the different nominal battery voltage would be transparent to the controller.

Perhaps, I don't know.

(09-05-2015 01:52 AM)matthiaspaul Wrote:  Don't they both run at 75 MHz by default? If so, keyboard debounce timing should be the same.

The 49G+ was famous for very bouncy keys. Even setting up the KEYTIME properly it was common to get ddoubble keystrokes. The main reason for the 50g to exist was to fix this keyboard problem. But I believe newRPL will not suffer from that, even on an old 49G+.

[Claudio L.]

(09-05-2015 07:48 AM)Zweb Wrote:  Good news!
I've translated and posted it on cnCalc (an Chinese calculator forum): http://www.cncalc.org/thread-11707-1-1.html , and I will translate the feedback to English and post here(if any).

Thank you!

[DavidM]

OK, so I installed the new firmware (twice) on my 49g+. Haven't done much yet, but here's some observations so far:

- The first attempt to install appeared to succeed, but I noticed right away that the calculator wouldn't power up with the "ON" key. I had to remove a battery for a couple seconds and replace it to get a response. Likewise, turning the calc off didn't happen when I pressed RS ON, either. The screen blanks for a brief moment, then immediately redraws with a "Memory Recovered" message at the bottom.

- Out of curiosity, I tried holding the RS and ON keys down longer than usual, thinking a key timing issue might be the cause. In that situation, the calculator turns off, but won't turn back on unless I do a PCR (paper clip reset).

- After trying several combinations of things, I noticed that in an earlier post you mentioned the flashing IO annunciator. That annunciator was not flashing on my unit, at least not at a humanly visible rate. It seemed slightly more dim than usual, so I suspect it was actually flashing very rapidly. Even with this condition, numeric input and basic math operations (+ - * /) seemed to work normally. But I went ahead and reflashed the newRPL firmware to see if the above anomalies would go away.

- After the reflash, the IO annunciator is now flashing as expected. However, the power on/off anomalies mentioned above still exist in the same form. So for now, my "power key" is simply to remove or replace a battery.

I'll play with it some more as I have a moment over the next few days.

[Claudio L.]

(09-05-2015 03:00 PM)DavidM Wrote:  OK, so I installed the new firmware (twice) on my 49g+. Haven't done much yet, but here's some observations so far:

- The first attempt to install appeared to succeed, but I noticed right away that the calculator wouldn't power up with the "ON" key. I had to remove a battery for a couple seconds and replace it to get a response. Likewise, turning the calc off didn't happen when I pressed RS ON, either. The screen blanks for a brief moment, then immediately redraws with a "Memory Recovered" message at the bottom.

This is expected. The 49g+ required paperclip reset after flashing. The 50g added the option of pressing Enter.
The off function does a warmstart for now.

(09-05-2015 03:00 PM)DavidM Wrote:  - After the reflash, the IO annunciator is now flashing as expected. However, the power on/off anomalies mentioned above still exist in the same form. So for now, my "power key" is simply to remove or replace a battery.

I'll play with it some more as I have a moment over the next few days.

In theory there's no reason to require a second flash, but I observed that too on mine. Some bytes were not flashed correctly. In my case I blamed the batteries and put new ones but perhaps if the rom hasn't been flashed in a long time some bits may refuse to be erased or reprogrammed. The second burn with the same bit states seems to do it. Now that I'm reflashing very often the issue never reappeared.

[DavidM]

(09-05-2015 09:19 PM)Claudio L. Wrote:  This is expected. The 49g+ required paperclip reset after flashing. The 50g added the option of pressing Enter.

I probably should have been more specific about what I meant. Even the PCR wouldn't "wake" the calculator; I had to remove/insert a battery to get the system to respond.

(09-05-2015 09:19 PM)Claudio L. Wrote:  The off function does a warmstart for now.

Good to know! Did I miss that in the docs somewhere? I don't recall seeing it. It's OK if it's not (I understand the current status of things), but I just wanted to make sure I'm not missing something.

(09-05-2015 09:19 PM)Claudio L. Wrote:  ...Some bytes were not flashed correctly. In my case I blamed the batteries and put new ones but perhaps if the rom hasn't been flashed in a long time some bits may refuse to be erased or reprogrammed. The second burn with the same bit states seems to do it. Now that I'm reflashing very often the issue never reappeared.

Is it possible that the status of the RTC could be a part of this? Or possibly something in RAM (not flash)? I don't recall ever seeing this type of issue with other firmware updates. Just a thought.

[Claudio L.]

(09-06-2015 04:39 PM)DavidM Wrote:  Good to know! Did I miss that in the docs somewhere? I don't recall seeing it. It's OK if it's not (I understand the current status of things), but I just wanted to make sure I'm not missing something.

Not on the docs, just above in this thread.

(09-06-2015 04:39 PM)DavidM Wrote:  Is it possible that the status of the RTC could be a part of this? Or possibly something in RAM (not flash)? I don't recall ever seeing this type of issue with other firmware updates. Just a thought.

Much of the interaction between the boot loader and the actual firmware is undocumented. I disassembled the start of the boot loader after a reset and there is very minimal configuration before passing control to the firmware, so it should never crash after a reset. At least in theory. NewRPL tries to reconfigure everything as if the bootloader wasn't there. But there might be unknown hardware wirings that require specific configurarion, it's still an unfinished task.

[Helix]

I seem to remember that in the previous newRPL rom I tested, the calculator did not start after the memory flash, and I had to remove a battery and insert it again in order to turn the system on. This time the calculator started nicely just after the update, perhaps because my previous rom was already a newRPL file?

I've found a bug: entering the letter E returns 0, instead of 'E' if there is no variable 'E' or the content of E. A following operation on the stack causes a system error ("exception: data abort").

I have also a problem with the gray menus. If I want a reasonable contrast to see the variable names, I have to set the contrast (ON +, ON -) to make these menus almost "white". I think this could work (although I have tested this new version only under artificial lightning), but the alpha indicator is then barely visible.

Like other members, I have to read the docs now.

[Claudio L.]

(09-06-2015 08:01 PM)Helix Wrote:  I seem to remember that in the previous newRPL rom I tested, the calculator did not start after the memory flash, and I had to remove a battery and insert it again in order to turn the system on. This time the calculator started nicely just after the update, perhaps because my previous rom was already a newRPL file?

I saw several variables in RAM that the firmware uses to interact with the bootloader. Perhaps the stock rom leaves something behind that the bootloader clears on the second reset.

(09-06-2015 08:01 PM)Helix Wrote:  I've found a bug: entering the letter E returns 0, instead of 'E' if there is no variable 'E' or the content of E. A following operation on the stack causes a system error ("exception: data abort").

I didn't get a chance to test this but this is a real bug I need to investigate and fix.

(09-06-2015 08:01 PM)Helix Wrote:  I have also a problem with the gray menus. If I want a reasonable contrast to see the variable names, I have to set the contrast (ON +, ON -) to make these menus almost "white". I think this could work (although I have tested this new version only under artificial lightning), but the alpha indicator is then barely visible.

Like other members, I have to read the docs now.

I agree it's almost white now. I'm still experimenting with the shades. It currently uses a 6 (in a 0 to 15, where 15 is black). I tried a darker grey (8) but didn't like the contrast. Perhaps it should be black on white rather than grey.

[Han]

(09-05-2015 10:45 AM)Claudio L. Wrote:  

(09-05-2015 01:52 AM)matthiaspaul Wrote:  I don't own a 49g+ so I don't know for sure, but since the ADC won't be able to cope with the full battery voltage anyway, the hardware will most probably contain some kind of voltage divider to bring down the maximum possible voltage to an acceptable voltage for the analog input. If so, the hardware developers might have adjusted the voltage divider so that the different nominal battery voltage would be transparent to the controller.

Perhaps, I don't know.

(09-05-2015 01:52 AM)matthiaspaul Wrote:  Don't they both run at 75 MHz by default? If so, keyboard debounce timing should be the same.

The 49G+ was famous for very bouncy keys. Even setting up the KEYTIME properly it was common to get ddoubble keystrokes. The main reason for the 50g to exist was to fix this keyboard problem. But I believe newRPL will not suffer from that, even on an old 49G+.

Just tried to flash an HP49G+ and the flashing process went without a hitch. However, after the update is done the screen shows: WAIT RESET

After waiting for a prolong period, I just used the reset pin... and saw a blank screen (no sign that it even turned on). So I reflashed it back to 2.15. I wonder if I should have tinkered with the contrast setting... will try again later.

[Guenter Schink]

(09-08-2015 05:09 PM)Han Wrote:  Just tried to flash an HP49G+ and the flashing process went without a hitch. However, after the update is done the screen shows: WAIT RESET

After waiting for a prolong period, I just used the reset pin... and saw a blank screen (no sign that it even turned on). So I reflashed it back to 2.15. I wonder if I should have tinkered with the contrast setting... will try again later.

As reported earlier, same happened to me with a 50G. The second attempt worked though. I guess best is to immediately do a reset, once the flashing was successful. If that fails too, removing and reinserting a battery might help.

Günter

[Han]

Here were the exact steps I took to get mine to work:

Flashed new firmware. When the screen shows WAIT RESET OR ENTER, I pressed ENTER. Then I had to remove all power sources (battery and USB cable -- I tend to use USB power when possible). Then I restored power (can either be putting batteries back in or just plug in the USB cable). I was then shown the prompt to install into RAM via "Cont" in the menu. The display looked like it was slightly garbled. Turning it off and back on seemed to do the trick. Thanks to Günter -- I did not realize what you had meant in your last post, but now it makes more sense.

With the USB cable plugged in, one cannot turn the calculator off. It immediately turns back on.

I am wondering if I need to repeat the steps above on the HP49G+. Will test tomorrow.

EDIT: (Comments on the demo)

The variable keys should not activate the actual variables unless the VARS menu is visible. Right now, pressing APPS will recall the first variable even when the VARS menu is not visible. I would imagine that this key would open apps (whatever that may be). Perhaps a regular press of VARS should show the VARS menu in the "dark" menus whereas long-pressing VARS should show the "light-grey" menus. So long as one of them is in view, then the corresponding keys (F1-F6 or G-L) should operate as soft menu keys. The idea I had in mind is that the "dark" menus are used to navigate the directory of various commands like back on the HP48 series. However, long-pressing VARS would bring up the VARS menu without affecting the dark menus so that one has access to both the directory of commands as well as one's own variables. On the other hand, if one simply expects the same behavior as that of the HP48 series, then a regular keypress would suffice.

I like the font, except that 'w' and 'm' are one pixel wider than the rest.