WP 34S --> WP 31S

[Jeff O.]

(03-28-2014 01:10 PM)walter b Wrote:  But I don't agree about taking politics out of discussion - IMHO it's in fact a topic which should be discussed more frequently and widely, just not here.

Definitely not here, but in general I agree that discussion can be good. Unfortunately, what passes for political discussion, at least on this (the west) side of the Atlantic, is just simultaneous shouting with no listening or (Heaven forbid) attempt to understand the other side.

(03-28-2014 01:10 PM)walter b Wrote:  Anyway, is anybody going to contact Eric?

d:-)

Happily, Eric has joined the conversation below.

[jebem]

(03-28-2014 05:17 PM)Jonathan Cameron Wrote:  Yes, this is a new project which we have almost completed our initial version. If you would like to beta test it on a windows emulator or to get the flash image to try it on a 30b device, you can visit the 31s branch on the sourceforge page:

http://sourceforge.net/p/wp34s/code/HEAD...hes/wp31s/

-Jonathan

Thank You, Jonathan, for the enlightenment!
As I said, I'm building one WP-34S (just received the HP-30B today!) but unfortunately I guess I will have to wait a few weeks to receive the materials from China to build the USB Serial cable. In the meantime I can use Windows or Linux (I'm a VMware consultant and VCI) to do some beta testing in my spare time.

[Paul Dale]

(03-28-2014 05:05 PM)jebem Wrote:  Is this 31S a new project, or are we just brainstorming an hypothetical RPN calculator?

As Jonathan already mentioned, it is a new project. It is based on the 34S code base so all the numerics will be the same. Well almost, I've added the octant reduction to the trig code so that SIN(180+x) = -SIN(180-x) e.g. A side effect of this will be a performance increase for some arguments.

I'd also like to move the statistical distributions from keystroke programs back into native C -- faster predominantly but also potentially a little more accurate. I don't have a time frame for completing this and it will require a fair bit of stabilisation.

I'd also like to try to find a faster algorithm for logarithms which are a bit of a sore point on the 34S.


- Pauli

[Jonathan Cameron]

(03-28-2014 11:01 PM)jebem Wrote:  In the meantime I can use Windows or Linux (I'm a VMware consultant and VCI) to do some beta testing in my spare time.

Jose,

It would be great if you could test the QtGui version. I've created a build that you can download from my site:

http://jmcameron.net/wp/wp31s_qtgui-2014-03-28.tar.gz

This is a 64-bit build done on Ubuntu 12.04. It may work on other Linux 64-bit builds with appropriate Qt libraries installed. You can get an idea of about the Qt issues from this README file: http://sourceforge.net/p/wp34s/code/HEAD..._QTGUI.txt

I added a dialog that shows the keyboard shortcuts for the QtGui version (under the QtGui help menu):



Note that many of these shortcuts are adapted from the Windows emulator, but there are a number of differences. The windows gui can use the Control key for the [f] shift, but the QtGui emulator is not set up for that and uses the Tab key instead.

The tar file contains a brief README file that tells how to install it.

Enjoy!

-Jonathan

[jebem]

Thank You, Jonathan!
I will have a try, once I setup one VM running Ubuntu.
By the way, thanks for your and the other contributors posts on the serial cable building thread here (very useful info - I ordered the components, just waiting for delivery now...):
http://www.hpmuseum.org/forum/thread-183...gle+switch

[Jonathan Cameron]

(03-29-2014 09:32 AM)jebem Wrote:  Thank You, Jonathan!
I will have a try, once I setup one VM running Ubuntu.
By the way, thanks for your and the other contributors posts on the serial cable building thread here (very useful info - I ordered the components, just waiting for delivery now...):
http://www.hpmuseum.org/forum/thread-183...gle+switch

You are welcome! If you figure out how to get the cable working under Linux, please let me know. I'm tried of switching over to Windows to flash new images!

I'm looking forward to some images of your cable.

I'm open to suggestions about the QtGui keyboard shortcuts and I hope you enjoy it.

-Jonathan

[jebem]

(03-29-2014 09:38 AM)Jonathan Cameron Wrote:  You are welcome! If you figure out how to get the cable working under Linux, please let me know. I'm tried of switching over to Windows to flash new images!
I'm looking forward to some images of your cable.
I'm open to suggestions about the QtGui keyboard shortcuts and I hope you enjoy it.
-Jonathan

Thanks for your support!
Ok, I have just created one VM with Ubuntu 12.04 and have copied the wp31s to it.
The QTgui calculator seems to run normally under Ubuntu, now it takes time to play with it Is there any test procedures to follow, or one can just play at will?

I can't wait to have the components delivered so I can start building the cable and start doing some testing.

[jebem]

(03-28-2014 11:36 PM)Paul Dale Wrote:  As Jonathan already mentioned, it is a new project. It is based on the 34S code base so all the numerics will be the same. Well almost, I've added the octant reduction to the trig code so that SIN(180+x) = -SIN(180-x) e.g. A side effect of this will be a performance increase for some arguments.

I'd also like to move the statistical distributions from keystroke programs back into native C -- faster predominantly but also potentially a little more accurate. I don't have a time frame for completing this and it will require a fair bit of stabilisation.

I'd also like to try to find a faster algorithm for logarithms which are a bit of a sore point on the 34S.

- Pauli

Thanks for your information, Pauli!
I'm preparing myself to build a WP-34S (late arrival...) and was curious about this new 31S project. Maybe the 34S project can benefit from some of the improved algorithms created here for the 31S?

[Paul Dale]

The 34S lacks the free memory space for the improvements, so they are unlikely to be ported across.


- Pauli

[Jonathan Cameron]

(03-29-2014 10:30 PM)jebem Wrote:  

(03-29-2014 04:07 AM)Jonathan Cameron Wrote:  It would be great if you could test the QtGui version.

Thanks for your support!
Ok, I have just created one VM with Ubuntu 12.04 and have copied the wp31s to it.
The QTgui calculator seems to run normally under Ubuntu, now it takes time to play with it Is there any test procedures to follow, or one can just play at will?

Jose,

I do not have test procedure (yet), but I'm working on it.
I've just ported the ydisplay mods from the trunk (wp-34s) to the WP-31s branch (they are not commited yet). If you (or anyone else) would like to test the compiled QtGui executables, let me know by email and I'll make them available on my website.

-Jonathan (you can email me at gmail.com: jmcameron)

[Jonathan Cameron]

We have integrated Bit's ydisplay mod into the WP-31s code. Marcus has built a version with the following options enabled:

* INCLUDE_YREG_CODE (shows Y in the upper display area)
* PRETTY_FRACTION_ENTRY (switches to fraction display mode instantly when a second radix is pressed)
* RP_PREFIX (polar y-register prefix)

If you would like to try it, the windows gui and the flash image have been rebuilt at:

http://sourceforge.net/p/wp34s/code/HEAD...hes/wp31s/

We would like you to try this update and let us know if you encounter any issues with it, or just like it, please let us know!

If you want to test the QtGui version, I have a parallel build on 64-bit Ubuntu 12.04 that I can make available for you (PM me).

-Jonathan

[Jeff O.]

(04-02-2014 02:20 AM)Jonathan Cameron Wrote:  We have integrated Bit's ydisplay mod into the WP-31s code. Marcus has built a version ...

Excellent, I was hoping the y-display mode would be included soon.

(04-02-2014 02:20 AM)Jonathan Cameron Wrote:  We would like you to try this update and let us know if you encounter any issues with it, or just like it, please let us know!

Considered it tried, I like it! I am hopeful that Eric will produce overlays so this can go into production.

Only one comment so far on operation. Perhaps UNDO is a work in progress, but seeing as how this is the first appearance of such a function on an RPN calculator, I had to try it. I'm not quite sure if it operates exactly as I would have expected. For example, if I do 1 ENTER 2 +, then UNDO, I would hope to get a 1 in y and a 2 in x. Instead, I get a 1 back in both x and y. If I do 1 ENTER 2 swap swap (so that 1 is in y, 2 is in x and stack lift is enabled) + UNDO, I do get 1 in y and 2 in x. I can see that in the first case, prior to the + operation, technically there was a 1 in both x and y as the 2 had not been entered yet. But I believe that classically speaking, as soon as you hit the +, the 2 is supposed to be entered into the x register, so the state of the machine should be as in the second case, i.e. 1 in y, 2 in x and stack lift enabled. Hopefully my description of the actual and desired behavior is clear.

Please don't take the above as criticism, overall this is really great work.

[Jonathan Cameron]

(04-02-2014 01:44 PM)Jeff O. Wrote:  Considered it tried, I like it! I am hopeful that Eric will produce overlays so this can go into production.

Only one comment so far on operation. Perhaps UNDO is a work in progress, but seeing as how this is the first appearance of such a function on an RPN calculator, I had to try it. I'm not quite sure if it operates exactly as I would have expected. For example, if I do 1 ENTER 2 +, then UNDO, I would hope to get a 1 in y and a 2 in x. Instead, I get a 1 back in both x and y.

Jeff,

Thanks for trying it.

Good catch on the UNDO. Pauli implemented the UNDO; perhaps he will chime in here.

-Jonathan

[Paul Dale]

Guess I didn't think UNDO through fully

The analysis is correct. In the first instance UNDO is undoing the + operation which includes the command line parsing and X replacement. In the second, it is still undoing the + operation but here there is no command line parsing taking place.

Although the current behaviour is surprising, it seems like the right thing to do in light of the case of 1 ENTER UNDO. The ENTER clearly must be undone completely. Likewise, 1 FILL UNDO.


- Pauli

[Jeff O.]

(04-02-2014 09:47 PM)Paul Dale Wrote:  Although the current behaviour is surprising, it seems like the right thing to do in light of the case of 1 ENTER UNDO is considered. The ENTER clearly must be undone completely. Likewise, 1 FILL UNDO.

I'm not sure how I think it should be done. Your way seems to offer some advantages and might seem quite natural once it is gotten used to.

[Paul Dale]

I suspect we'll get some suggestions as to how UNDO should behave....


- Pauli

[walter b]

(04-03-2014 08:42 AM)fhub Wrote:  What has happened to the WP34s/WP31s code section on Sourceforge?
http://sourceforge.net/p/wp34s/code/

All files have gone ...

It looked ok yesterday. Strange.

d:-?

[walter b]

That's what I've written in the manual about UNDO:

If you have erroneously called a wrong command, just press UNDO to undo it. This command will restore the stack and all other registers exactly as they were before the last command was executed.

Now we only have to talk about what is a command and what's not. The usual monadic, dyadic, etc. functions are easy. A plain numeric entry (pressing some digits, even including radix marks and EEX) isn't a command IMHO.

Addenda:
What about <- ? Shall be no command unless executed as CLx.
Sigma+ ? Shall be reversible by UNDO though there's no pressing need since Sigma- exists.
ENTER, FILL, CLALL, RESET ? Shall be commands.

d:-)

[Jonathan Cameron]

(04-03-2014 09:24 AM)walter b Wrote:  

(04-03-2014 08:42 AM)fhub Wrote:  What has happened to the WP34s/WP31s code section on Sourceforge?
http://sourceforge.net/p/wp34s/code/

All files have gone ...

It looked ok yesterday. Strange.

d:-?

I noticed that last night but was still able to commit a change (to the QtGui code in the 31s branch). And it it is still down this morning.

-Jonathan

[Jonathan Cameron]

(04-03-2014 09:24 AM)walter b Wrote:  

(04-03-2014 08:42 AM)fhub Wrote:  What has happened to the WP34s/WP31s code section on Sourceforge?
http://sourceforge.net/p/wp34s/code/

All files have gone ...

It looked ok yesterday. Strange.

It seems to be back now.

-Jonathan