Threaded Mode | Linear Mode

Bit · 01-31-2015, 01:32 AM

(01-31-2015 01:11 AM)BarryMead Wrote:
(11-27-2014 02:13 AM)Bit Wrote: Pascal, I've changed the function names as you requested.

The new NEXTP implementation still got stuck in a loop for positive infinity in double precision mode, I've fixed that, too.

Bit:

Should we expect a new build in the "Bits 34s and 31S patches and custom binaries" thread soon?
If not there where will the new stuff be posted?

Yes, I've been working on some improvements to angular conversions that I'm planning to publish soon. But I don't understand why you quoted my post about NEXTP. My builds add the PREVP command but that's it, all fixes to NEXTP have been incorporated into the official version.

BarryMead · 01-31-2015, 01:34 AM

(01-31-2015 01:32 AM)Bit Wrote: Yes, I've been working on some improvements to angular conversions that I'm planning to publish soon. But I don't understand why you quoted my post about NEXTP. My builds add the PREVP command but that's it, all fixes to NEXTP have been incorporated into the official version.

I just replied to the last posting in the thread. Sorry about the confusion.

Bit · 02-07-2015, 06:17 PM

A few problems with rounding in significant figures display mode have been fixed in the latest SVN commit that also introduced a new user interface instead of flags.

The official builds are unaffected because significant figures mode is disabled by default.

MarkHaysHarris777 · 02-08-2015, 02:17 PM

Which build is recommended as 'stable,' and which build(s) are considered experimental (in regression testing)?

I am noticing that the build from last July is still on the front burner, so to speak, on the main download site... but I was able to find the version 3.3 3742 on source forge (I've run both, but alas, still am not comfortable enough with the 34s to really make an honest comparison).

Which versions are recommended?

thank you

Nigel (UK) · 02-08-2015, 03:07 PM

(02-08-2015 02:17 PM)MarkHaysHarris777 Wrote: Which build is recommended as 'stable,' and which build(s) are considered experimental (in regression testing)?

I am noticing that the build from last July is still on the front burner, so to speak, on the main download site... but I was able to find the version 3.3 3742 on source forge (I've run both, but alas, still am not comfortable enough with the 34s to really make an honest comparison).

Which versions are recommended?

thank you

It isn't that simple - there are multiple versions of each build depending on which compile-time options are set! I'd recommend downloading one of Bit's builds - they turn on some extra features, and Bit seems to have an inhuman skill for finding bugs. You'll find them in http://www.hpmuseum.org/forum/thread-2433.html.

Remember that new builds are a major part of the fun of having a WP34S!

Nigel (UK)

walter b · 02-08-2015, 04:44 PM

(02-08-2015 02:17 PM)MarkHaysHarris777 Wrote: Which build is recommended as 'stable,' and which build(s) are considered experimental (in regression testing)?

I am noticing that the build from last July is still on the front burner, so to speak, on the main download site... but I was able to find the version 3.3 3742 on source forge (I've run both, but alas, still am not comfortable enough with the 34s to really make an honest comparison).

Depends. If you see a good chance you might need answers from the community shortly, taking the official build is recommended. 3742 should do. Else do as you please but don't expect much support. Your choice.

d:-)

Bit · 02-08-2015, 05:58 PM

(02-08-2015 04:44 PM)walter b Wrote: Depends. If you see a good chance you might need answers from the community shortly, taking the official build is recommended. 3742 should do. Else do as you please but don't expect much support. Your choice.

If someone decides to use my builds, they should read the introductory post as it explains every single change that's been made and some of the reasons behind them, too. Most changes aren't incompatible with the official builds in the sense that they either only add new functionality or only change the UI and won't break programs. There are some subtle exceptions one might run into if using programming patterns that I'd consider rather untypical. I'm happy to answer questions and fix any issues found, but as I'm just one person, I obviously can't guarantee that I'll be as quick to respond as a whole community together.

Nigel mentioned that I had fixed a lot of bugs. Everything that was clearly a bugfix rather than an enhancement has also been committed to the official builds.

Bit · (This post was last modified: 02-20-2015 05:06 AM by Bit.)

The gradian symbol was not handled properly in the Qt emulator with 'real fonts' enabled. The problem was caused by set_status_sized() in display.c interpreting the second '\007' incorrectly in escape sequences. I've committed a fix.

Thanks to Barry for pointing out the issue.

Paul Dale · 02-20-2015, 05:53 AM

Not a bug, but an improvement nonetheless.

I finally replaced three of the quantile functions with Dieter's improved versions last night. This should have an impact on speed and accuracy for both the 31S and 34S. This will be most marked for the χ² quantile function -- previously it used a generic Newton based solver, now it is using a custom solver with cubic convergence. The other two quantile functions modified are Student's T and the normals.

I've still got an accuracy improvement for the Student's T CDF to put in, hopefully this weekend.

That just leaves Snedecor's F, Poisson and Binomial using the generic solver.

- Pauli

pascal_meheut · 02-20-2015, 08:28 AM

(02-20-2015 04:52 AM)Bit Wrote: The gradian symbol was not handled properly in the Qt emulator with 'real fonts' enabled. The problem was caused by set_status_sized() in display.c interpreting the second '\007' incorrectly in escape sequences. I've committed a fix.

I've built and uploaded new versions of the Qt Emulator to include this fix (and Paul's one).

Paul Dale · 02-20-2015, 10:19 AM

(02-20-2015 08:28 AM)pascal_meheut Wrote: I've built and uploaded new versions of the Qt Emulator to include this fix (and Paul's one).

Mine isn't finished yet.......

pascal_meheut · 02-20-2015, 11:44 AM

(02-20-2015 10:19 AM)Paul Dale Wrote:
(02-20-2015 08:28 AM)pascal_meheut Wrote: I've built and uploaded new versions of the Qt Emulator to include this fix (and Paul's one).

Mine isn't finished yet.......

No problem, I'll build new ones once it is finished.

Bit · 02-24-2015, 03:50 AM

The Qt 82240B emulator only printed a single pixel of spacing between characters instead of two as the real device. That's how 24 characters, each 5 pixel wide, use up the entire horizontal resolution: 24*5 + 23*2 = 166.

The following is perfectly aligned when printed by the real printer but not in the emulator:
XXX
XΔX

I've committed a fix to SVN.

pascal_meheut · 02-24-2015, 06:42 AM

I've built a new version and uploaded it to sourceforge.

Bit · 03-15-2015, 04:55 PM

Dieter reported that the DISP command didn't work correctly in the two significant figures modes. The problem was that arguments >= 8 were accepted. I've committed a fix.

Dieter · (This post was last modified: 03-15-2015 10:42 PM by Dieter.)

(03-15-2015 04:55 PM)Bit Wrote: Dieter reported that the DISP command didn't work correctly in the two significant figures modes. The problem was that arguments >= 8 were accepted.

Why was this a problem? DISP 9 works fine in all display modes other than SIG and SIG0.

(03-15-2015 04:55 PM)Bit Wrote: I've committed a fix.

Does the fix now reject arguments ≥8 so that e.g. DISP 9 is no longer allowed?

Dieter

Bit · 03-15-2015, 10:50 PM

(03-15-2015 10:41 PM)Dieter Wrote: Does the fix now reject arguments ≥8 so that e.g. DISP 9 is no longer allowed?

The SIG and SIG0 display modes only support arguments between 0 and 7. DISP now rejects arguments >= 8 only when in SIG or SIG0 mode, other modes are unaffected.

emece67 · (This post was last modified: 05-06-2015 10:02 AM by emece67.)

Not sure if this is a bug or other thing, but as I've see it first time when switching from yagarto 4.6.0 to Atmel 4.9.3, I'm wandering if this is an unknown (to me) feature or a problem related to the compiler.

What means that funny "4dr" in the exponent field?

[Image: funny_symb.jpg?dl=1]

p.s. I'm aware that switching from yagarto 4.6.0 is not the correct way, but such 4000 more free flash words seem irresistible...

Massimo Gnerucci · 05-06-2015, 11:46 AM

(05-06-2015 09:59 AM)emece67 Wrote: Not sure if this is a bug or other thing, but as I've see it first time when switching from yagarto 4.6.0 to Atmel 4.9.3, I'm wandering if this is an unknown (to me) feature or a problem related to the compiler.

(...)

p.s. I'm aware that switching from yagarto 4.6.0 is not the correct way, but such 4000 more free flash words seem irresistible...

I think that's how you get 4000 more flash words... Smile

Marcus' post

BarryMead · 05-06-2015, 09:53 PM

(05-06-2015 09:59 AM)emece67 Wrote: Not sure if this is a bug or other thing, but as I've see it first time when switching from yagarto 4.6.0 to Atmel 4.9.3, I'm wandering if this is an unknown (to me) feature or a problem related to the compiler.

What means that funny "4dr" in the exponent field?

From numerous posts I have read on this forum, I firmly believe that using any compiler other than the yagarto 4.6.0 will cause mysterious errors such as the one you illustrated. I hope you kept your copy of yagarto 4.6.0 so you can restore error free compilation.

Best Wishes, Barry