View previous topic :: View next topic |
Author |
Message |
| angusj Site Admin
| Joined: 18 Jun 2005 | Posts: 406 | : | | Items |
|
Posted: Wed Aug 24, 2005 10:18 pm Post subject: |
|
|
Simes wrote: | rfb wrote: | Although you are improving generation times Angus seems to generate new puzzles almost instaneously. | Yes, I wonder what the trick is? |
Compile puzzles in a background thread and have a few spares ready to go. |
|
Back to top |
|
|
| Simes
| Joined: 08 Apr 2005 | Posts: 71 | : | Location: North Yorkshire, UK | Items |
|
Posted: Mon Oct 17, 2005 6:02 am Post subject: Version 2.4 now released |
|
|
Version 2.4, released today, has these improvements:
Implemented optical character recognition (OCR). It's not perfect, but it does a reasonable job on many images. Simply open, or drag and drop an image file containing Sudoku puzzles, or you can also paste images from the clipboard. If the OCR finds a single puzzle, it will be imported. If the image contains several puzzles, you can select the one you want to import. It the OCR doesn't find a puzzle, you can help it by selecting the puzzle within the image - this is usually when the puzzle forms only a small part of the overall image. The OCR isn't perfect, so please check the imported puzzle against the original image.
Added a new number selector. For those who prefer to use the mouse, this is much slicker than the pop-up menus. Left click for a big number, and right-click for a pencil mark. For those that prefer the pop-up menu, the selector can be disabled from the Options dialog. And of course, the keyboard can still be used.
Added a menu option to save the puzzle as an image file - PNG, GIF, JPG or BMP formats are supported.
Added a menu option to copy the puzzle to the clipboard as an image.
The "effects" display now reflects the current state of the pencil marks, rather than the simple effects of numbers already present. This means the effects are updated as you eliminate cell candidates.
When opening a file with "auto-calculate pencil marks" enabled, the pencil marks are only calculated if none are already present. This prevents any manually adjusted pencil marks in the file from being overwritten by the auto-calculated ones.
It can be downloaded from http://www.simes.clara.co.uk/programs/sudoku-setup.exe
Simon _________________ Simes
www.sadmansoftware.com/sudoku |
|
Back to top |
|
|
| rfb
| Joined: 09 Apr 2005 | Posts: 26 | : | | Items |
|
Posted: Mon Oct 17, 2005 4:22 pm Post subject: I never win the hotel vouchers :( |
|
|
The new version is nice - I don't know whether I will use the OCR facility but the selector wheel is useful.
I have been completing the Daily Telegraph competititon Sudokus for several months but have never won anything
It may be that I have just been unlucky but I fear my entries pasted from SadMan have been rejected as invalid format.
The rules (posted at http://www.sudoku.org.uk/competitions.htm) require simple numeric format while your (and Angus' program) produce a pretty format with separators between boxes.
Would it be possible to have an alternative Copy to clipboard format that meets the Telegraphs requirements?
I know I could edit the data after pasting but I was born lazy!
(howevere I don't cheat by using autosolve and I try to avoid using pencil marks except in diabolicals). _________________ Bob |
|
Back to top |
|
|
| Simes
| Joined: 08 Apr 2005 | Posts: 71 | : | Location: North Yorkshire, UK | Items |
|
Posted: Mon Oct 17, 2005 8:28 pm Post subject: |
|
|
Hi Bob,
Perhaps you've just been unlucky.
Anyway, to rule out disqualification for not following the submission format, I'm currently in the process of adding options for specifying the format to be used.
I'll let you have a pre-release version when it's ready - shouldn't be long.
Simon _________________ Simes
www.sadmansoftware.com/sudoku |
|
Back to top |
|
|
| rfb
| Joined: 09 Apr 2005 | Posts: 26 | : | | Items |
|
Posted: Sat Oct 29, 2005 12:14 pm Post subject: |
|
|
I am atill unlucky - no hotel vouchers yet!
A couple of suggestions for the number selector:
- Don't display selector on double click for only candidate, it uneccessary flicker
- Allow right click on selector number to toggle pencil mark, I am too used to using left click to move to a cell (this would allow first right click to be pop up menu again)
- Could the colours be added to the selector, possibly as an inner or outer ring
Bob _________________ Bob |
|
Back to top |
|
|
| Simes
| Joined: 08 Apr 2005 | Posts: 71 | : | Location: North Yorkshire, UK | Items |
|
Posted: Sat Oct 29, 2005 7:00 pm Post subject: |
|
|
The problem with not displaying the number selector on a double-click, is that a single click happens before the double-click. So the single click shows the selector, then a double-click happens and makes it go away. Having said that, some judicious use of a timer seems to do the trick.
I've also added colours to the selector.
I'm not sure about the left/right click of the selector to choose whether it's a bug number/pencil mark is entered. I think people have already made that choice before the selector is shown.
I'll send you a new version shortly.
Simon _________________ Simes
www.sadmansoftware.com/sudoku |
|
Back to top |
|
|
| Pat
| Joined: 06 Sep 2006 | Posts: 128 | : | | Items |
|
Posted: Thu Dec 28, 2006 12:17 pm Post subject: SadMan - fails to find the Swordfish |
|
|
hey Simes, are you still here? long time silent -- i've just downloaded your software (version 2.8.1)
and tried it on this puzzle [33 clues] -
Code: |
6 . 7 | 3 . 8 | 5 . 4
5 . . | 4 . 6 | . . 9
. . . | . 1 . | . . .
-------+-------+------
9 6 . | . . . | . 4 8
. . 4 | . 6 . | 1 . .
7 1 . | . . . | . 9 6
-------+-------+------
. . . | . 3 . | . . .
4 . . | 5 . 7 | . . 1
3 . 6 | 1 . 4 | 9 . 7
|
- with solving-techniques enabled up to Swordfish (inclusive),
it reaches this point [49] -
Code: |
+---+---+---+
|6.7|3.8|514|
|5.1|476|8.9|
|.4.|.15|...|
+---+---+---+
|96.|.51|.48|
|..4|.6.|1..|
|715|84.|.96|
+---+---+---+
|17.|63.|4..|
|4..|5.7|..1|
|356|1.4|9.7|
+---+---+---+
+-------------------+-------------------+-------------------+
| 6 29 7 | 3 29 8 | 5 1 4 |
| 5 23 1 | 4 7 6 | 8 23 9 |
| 28 4 2389 | 29 1 5 | 67 67 23 |
+-------------------+-------------------+-------------------+
| 9 6 23 | 27 5 1 | 237 4 8 |
| 28 238 4 | 279 6 239 | 1 357 35 |
| 7 1 5 | 8 4 23 | 23 9 6 |
+-------------------+-------------------+-------------------+
| 1 7 289 | 6 3 29 | 4 258 25 |
| 4 289 28 | 5 289 7 | 36 36 1 |
| 3 5 6 | 1 28 4 | 9 28 7 |
+-------------------+-------------------+-------------------+
|
- and fails to find the Swordfish (r129 2 in c258)
~ Pat |
|
Back to top |
|
|
| Simes
| Joined: 08 Apr 2005 | Posts: 71 | : | Location: North Yorkshire, UK | Items |
|
Posted: Thu Dec 28, 2006 12:46 pm Post subject: |
|
|
Hi Pat,
Yes, still here.
That is interesting. I'm just on my way out - I'm away for new year - but I'll take a closer look next week.
Thanks for letting me know. _________________ Simes
www.sadmansoftware.com/sudoku |
|
Back to top |
|
|
| Simes
| Joined: 08 Apr 2005 | Posts: 71 | : | Location: North Yorkshire, UK | Items |
|
Posted: Tue Jan 02, 2007 5:32 pm Post subject: |
|
|
Thanks Pat,
I've identified and fixed the problem and will be releasing the next version in a few days. _________________ Simes
www.sadmansoftware.com/sudoku |
|
Back to top |
|
|
| Pat
| Joined: 06 Sep 2006 | Posts: 128 | : | | Items |
|
Posted: Fri Jan 05, 2007 8:38 am Post subject: re: SadMan version 2.9 |
|
|
that's great, Simes, thanks!!
so, what next?
let me be greedy -- how about Jellyfish? |
|
Back to top |
|
|
| Pat
| Joined: 06 Sep 2006 | Posts: 128 | : | | Items |
|
Posted: Thu May 10, 2007 10:50 am Post subject: SadMan - fails to find the Block / Block interactions |
|
|
trying SadMan SuDoku version 2.9.1
on this puzzle [20 clues] -
Code: |
. . . | 3 . 8 | . . .
. . 6 | . 4 . | 1 . .
. 8 . | . . . | . 9 .
-------+-------+------
4 . . | . . . | . . 5
. 7 . | . . . | . 8 .
1 . . | . . . | . . 3
-------+-------+------
. 6 . | . . . | . 7 .
. . 4 | . 5 . | 6 . .
. . . | 7 . 3 | . . .
|
- with only "singles" + "Block / Block Interactions" enabled,
it reaches this point [27] -
Code: |
+---+---+---+
|.4.|3.8|...|
|..6|.47|1.8|
|.8.|...|.9.|
+---+---+---+
|4..|...|..5|
|67.|...|.8.|
|1..|...|..3|
+---+---+---+
|.6.|...|.7.|
|7.4|85.|6..|
|...|763|...|
+---+---+---+
+----------------------+----------------------+----------------------+
| 259 4 12579 | 3 129 8 | 257 256 267 |
| 2359 2359 6 | 259 4 7 | 1 235 8 |
| 235 8 12357 | 1256 12 1256 | 23457 9 2467 |
+----------------------+----------------------+----------------------+
| 4 239 2389 | 1269 123789 1269 | 279 126 5 |
| 6 7 2359 | 12459 1239 12459 | 249 8 1249 |
| 1 259 2589 | 24569 2789 24569 | 2479 246 3 |
+----------------------+----------------------+----------------------+
| 23589 6 123589 | 1249 129 1249 | 234589 7 1249 |
| 7 1239 4 | 8 5 129 | 6 123 129 |
| 2589 1259 12589 | 7 6 3 | 24589 1245 1249 |
+----------------------+----------------------+----------------------+
|
- and fails to find the "Block / Block interactions" (c9 6 in box 3; or equally, r1 6 in box 3)
~ Pat |
|
Back to top |
|
|
| Simes
| Joined: 08 Apr 2005 | Posts: 71 | : | Location: North Yorkshire, UK | Items |
|
Posted: Thu May 10, 2007 6:06 pm Post subject: |
|
|
Hi Pat,
The interactions you identify are known to me as block/column/row interactions, and if you enable this technique, the puzzle will be solved.
I admit that it's a moot point though, since block/column/row and block/block are really just different ways of looking at the situation.
Simon _________________ Simes
www.sadmansoftware.com/sudoku |
|
Back to top |
|
|
| Pat
| Joined: 06 Sep 2006 | Posts: 128 | : | | Items |
|
Posted: Sun May 13, 2007 7:28 am Post subject: |
|
|
Simes wrote: |
The interactions you identify are known to me as block/column/row interactions, and if you enable this technique, the puzzle will be solved.
I admit that it's a moot point though, since block/column/row and block/block are really just different ways of looking at the situation.
|
yes, the puzzle can be solved with a "block/column/row interaction" -- but i deliberately disabled that -- the puzzle can be solved with a "Block / Block Interaction"
this is what i expect (using your terminology):Block/block interaction: blocks 6 and 9 have an interaction (this excludes the 6 at r1c8, leaving a "hidden single" in that row)
or it may have been in the other direction --Block/block interaction: blocks 1 and 2 have an interaction (this excludes the 6 at r2c9, leaving a "hidden single" in that column)
i'm sorry i must disagree with you but in my view the point is not moot -- the software indicates that the puzzle cannot be solved with "Block / Block Interactions" where in fact it can be.
additional examples of this situation:
see puzzles #4,8,10,11 in the topic Puzzles needing lots of locked candidates
~ Pat |
|
Back to top |
|
|
| Simes
| Joined: 08 Apr 2005 | Posts: 71 | : | Location: North Yorkshire, UK | Items |
|
Posted: Wed May 16, 2007 4:21 pm Post subject: |
|
|
Well you say the puzzle can be solved with block/block interaction, but the interactions you refer to are known to me as block/column/row interactions. Which is why the puzzle isn't solved with this technique disabled.
My distinction is simply that if the candidates in one of the blocks are entirely within a single row or column (wrt the other block), then it's BCR, if the candidates are in two columns or rows in both blocks then it's BB. It's only the candidates that are considered, not any big numbers. It's a bit arbitrary, but it works for me.
What I meant by "moot point" was that these two techniques are closely related, and one man's BB is another man's BCR. _________________ Simes
www.sadmansoftware.com/sudoku |
|
Back to top |
|
|
| Pat
| Joined: 06 Sep 2006 | Posts: 128 | : | | Items |
|
Posted: Sun May 20, 2007 7:33 am Post subject: re: Block / Block interactions |
|
|
OK, we'll agree to disagree.
~ Pat |
|
Back to top |
|
|
|