Sie sind vermutlich noch nicht im Forum angemeldet - Klicken Sie hier, um sich kostenlos anzumelden Impressum 
Sie können sich hier anmelden
Dieses Thema hat 89 Antworten
und wurde 3.490 mal aufgerufen
 More languages
Seiten 1 | 2 | 3 | 4 | 5 | 6
Raman Offline




Beiträge: 291

20.10.2026 13:32
#31 RE: 40 fun bonus tiles as in Anadrome Zitat · Antworten

You can at least answer these 2 questions:

☀ What is the highest theoretical possible score that Scrabble3D will be able to represent, with the data type that has been assigned to score variable? You can say the highest possible score whatever that can be achieved both without losing precision and with losing precision. You can also say for the latest version 4 of Scrabble3D and for the earlier version 3.1 of Scrabble3D. So, I ask this whenever I modify the Scrabble3D code to add more interesting bonus fields like power and product and if I do not switch over to big integer libraries. This has certainly also got nothing to do with the highest practically achievable score for each of the players either for Scrabble or for Super Scrabble under each of their own standard rules.

☀ Suppose I were to introduce random holes into the Scrabble board, while designing a newer board layout, then which of the files in the source code should I need to modify at least in the latest version 4 of Scrabble3D to instruct the computer that no tiles can be placed on those specific squares right from the very beginning of the game in play in the squares that have been marked as holes in the Scrabble board?

Zitat von Raman im Beitrag #29
Lux: 4.066×10117 (15100).
Raman: 5.281×10116 (6150).

For handling huge scores in C++, I think I am going to need big integer libraries like either GMP (GNU Multiple Precision) or Apfloat. The maximum value of either integer data type or long data type in C++ is 231-1 = 2147483647 and the minimum value is -231 = -2147483648. What will be the maximum value of score of any of the players that the latest version 4 of Scrabble3D that has been written in Qt6/Qml will be able to store? Is it 231-1 = 2147483647? What about the earlier version 3.1 of Scrabble3D that had been written in FreePascal? I actually just ask only theoretically. Even with the largest board size with 3D board in Scrabble3D, it would be impossible to achieve that score, at least with the Scrabble3D's default bonus fields.
In programming languages like Python and PARI/GP, it is possible to store integers that are thousands of digits long without losing precision. Data types like string, integer, float and boolean are also detected automatically. PARI/GP is a programming language that is designed especially for mathematics and specifically for number theory.

In programming languages like Java and Kotlin, the data types integer and long store smaller numbers without losing precision.
The range of integer is from -231 = -2147483648 to 231-1 = 2147483647.
The range of long is from -263 = -9223372036854775808 to 263-1 = 9223372036854775807.

The data types float and double store even longer numbers, but with losing precision. The precision of float is 7 decimal places and the precision of double is 16 decimal places.
The range of float is from -2128 = -3.403×1038 to 2128-1 = 3.403×1038.
The range of double is from -21024 = -1.798×10308 to 21024-1 = 1.798×10308.

At least in Java, there are also BigInteger and BigDecimal libraries. BigInteger data type can be at least be used to store integers that are thousands of digits long without losing precision. But, the BigInteger and BigDecimal libraries in Java are not well optimized and so they are much slower in performance than whenever compared to GMP (GNU Multiple Precision) and Apfloat libraries in C++. Although, it certainly does not make much of difference in tasks which will not need too much of computational power like calculating Scrabble scores, it will certainly have a lot of difference in computational time in rigorous tasks which will need too much of computational power like factoring larger integers.

When compared to data types like float and double in Java, there are also equivalent data types like single and double in C++. integer and long are almost equivalent data types both in C++ and in Java.

2 bytes are needed to store integer variables in C++ at least by using Turbo C++ compiler and so its range is from -215 = -32768 to 215-1 = 32767.
On the other hand, 4 bytes are needed to store integer variables in C++ at least by using GNU C++ compiler and so its range is from -231 = -2147483648 to 231-1 = 2147483647.

Finally, anyway, in any case, execution of C++ code will be always be relatively faster than execution of Java code, just because C++ compiler has been designed in such a way so as to translate C++ code directly into machine code. The BigInteger and BigDecimal libraries in Java are much slower in performance than whenever compared to GMP (GNU Multiple Precision) and Apfloat libraries in C++ because:
☀ They are not well optimized.
☀ Execution of C++ code will be always be relatively faster than execution of Java code, just because C++ compiler has been designed in such a way so as to translate C++ code directly into machine code.


WESPA Player Profile: Raman Viswanathan | Fast Word Study: Google Play Store and Indus App Store | Fast Word Quiz: Google Play Store and Indus App Store

Raman Offline




Beiträge: 291

20.10.2026 19:07
#32 RE: 40 fun bonus tiles as in Anadrome Zitat · Antworten

Zitat von Raman im Beitrag #31
In programming languages like Java and Kotlin, the data types integer and long store smaller numbers without losing precision.
The range of integer is from -231 = -2147483648 to 231-1 = 2147483647.
The range of long is from -263 = -9223372036854775808 to 263-1 = 9223372036854775807.

The data types float and double store even longer numbers, but with losing precision. The precision of float is 7 decimal places and the precision of double is 16 decimal places.
The range of float is from -2128 = -3.403×1038 to 2128-1 = 3.403×1038.
The range of double is from -21024 = -1.798×10308 to 21024-1 = 1.798×10308.
Actually, in programming languages like Java and Kotlin, if the difference in scores of 2 players is 1 point, then the integer data type will represent the winner properly as long as the winner's score is ≤ 231-1 = 2147483647. The long data type will represent the winner properly as long as the winner's score is ≤ 263-1 = 9223372036854775807.

The float data type will represent the winner properly as long as the winner's score is ≤ 224 = 16777216. The double data type will represent the winner properly as long as the winner's score is ≤ 253 = 9007199254740992.

In all of the above cases, if the winner's score exceeds the above mentioned range and if the loser's score is 1 lesser than the winner's score, then at the end of the game that had been played, the result should be always be reported as a tie. But, strangely, it looked like the Anadrome website picked up a random winner and reported that the winner won by 0 points, at the end of the game that had been played.

In the Anadrome website, quite often circumstances can actually arise only in 2 player games that could have been played over thereby, in which one of the players' score is 10100+3100 ≈ 1.000×10100 and in which another of the players' score is 10100+2100 ≈ 1.000×10100. The integer, long and float data types do not have the capacity to store up with numbers these big. It can be only stored up the double data type, but it always loses precision whenever trying out to store up with numbers these big and so both of those of above mentioned 2 numbers will be considered to be equivalent to each other, although, the 2 numbers will actually have a difference of as much as 3100-2100 ≈ 5.154×1047.


WESPA Player Profile: Raman Viswanathan | Fast Word Study: Google Play Store and Indus App Store | Fast Word Quiz: Google Play Store and Indus App Store

Scotty Offline

Administrator


Beiträge: 3.834

21.10.2026 07:31
#33 RE: 40 fun bonus tiles as in Anadrome Zitat · Antworten

Zitat von Raman im Beitrag #31
☀ What is the highest theoretical possible score that Scrabble3D

I think it's a signed integer resp. 2^16. But you need to check the code - and not only the calculation/s but also the feedback.

Zitat von Raman im Beitrag #31
☀ ... which of the files in the source code should I need to modify

Probably all ;-) The calculation of a move is done in computemove.cpp based on continuous fields. A simple solution could be to have a malus field with some high value, and to prevent placing words that would result in decrease in the total value.


Download: Version 4 | Help:Wiki | Discussion: Forum | News: Mastodon | Fanship: Facebook

Raman Offline




Beiträge: 291

22.10.2026 08:16
#34 RE: 40 fun bonus tiles as in Anadrome Zitat · Antworten

Okay. Thanks.

Zitat von Scotty im Beitrag #33
I think it's a signed integer resp. 2^16. But you need to check the code - and not only the calculation/s but also the feedback.
It looks like some compilers like Borland C++ compiler will use 16 bits for representing integer data type and so their range will be from -215 = -32768 to 215-1 = 32767, while other compilers like GNU C++ compiler will use 32 bits for representing integer data type and so their range will be from -231 = -2147483648 to 231-1 = 2147483647. In any case, it looks like Qt C++ compiler will always use 16 bits for representing integer data type and so its range will be always be from -215 = -32768 to 215-1 = 32767.
Zitat von Scotty im Beitrag #33
Probably all ;-) The calculation of a move is done in computemove.cpp based on continuous fields. A simple solution could be to have a malus field with some high value, and to prevent placing words that would result in decrease in the total value.
I can modify the code in any way so as to assign any mathematical function to be associated with the hole bonus field. For example, as an alternative to having a malus field with some high value to prevent placing words that would result in decrease in the total value, if any of tiles are placed on any of hole bonus fields, then I can also just return the negative of the score whatever the score that would be if no tiles were placed on any of hole bonus fields. This can at least stop the computer from playing any of tiles on any of hole bonus fields, as long as there are other places in Scrabble board to place tiles and the computer would better pass its turn if all of its score values by placing any of its tiles on hole bonus fields would be negative. On the other hand, human players, at least over a network game, can just play any of their tiles on hole bonus fields, either if they just do not understand what the hole bonus fields are or if they just do not want to follow the rules or if they just want to annoy me, even if those moves will just reduce their overall score and if they just do not want to play their own games for a win.

By the way, it looks like I can also modify the code in computemove.cpp to change the bingo bonus from 50 to 5×Number of tiles in the rack, as it had been in the Anadrome website.


WESPA Player Profile: Raman Viswanathan | Fast Word Study: Google Play Store and Indus App Store | Fast Word Quiz: Google Play Store and Indus App Store

Scotty Offline

Administrator


Beiträge: 3.834

22.10.2026 08:21
#35 RE: 40 fun bonus tiles as in Anadrome Zitat · Antworten

Zitat von Raman im Beitrag #34
It looks like I can also modify the code in computemove.cpp to change the bingo bonus from 50 to 5×Number of tiles in the rack, as it had been in the Anadrome website.

Be careful with this value. The bingo is considered at several other places too, as commented in the code.


Download: Version 4 | Help:Wiki | Discussion: Forum | News: Mastodon | Fanship: Facebook

Raman Offline




Beiträge: 291

22.10.2026 08:59
#36 RE: 40 fun bonus tiles as in Anadrome Zitat · Antworten

An English Scrabble board can be packed up to a maximum of 70%. I have tested this several times in Scrabble3D PC application in which we can customize with our own letter distribution.

Some of furthermore of facts about the Anadrome website:
Anadrome by default uses 109 tiles for a 15×15 board and has got no blank tiles. It goes on by ratio of number of occurrences of each letter in CSW2012 dictionary to number of tiles in bag rounded off to nearest integer.
https://github.com/atebits/Words.

Bingo bonus will always be 5×Number of tiles in rack and so for 7 tiles in rack, the bingo bonus will always be 35 points.

Anadrome by default uses 7 tiles in rack with 26 tiles in bag for 7×7 board.
Anadrome by default uses 7 tiles in rack with 33 tiles in bag for 9×9 board.
Anadrome by default uses 7 tiles in rack with 50 tiles in bag for 11×11 board.
Anadrome by default uses 7 tiles in rack with 77 tiles in bag for 13×13 board.
Anadrome by default uses 7 tiles in rack with 109 tiles in bag for 15×15 board.
Anadrome by default uses 7 tiles in rack with 140 tiles in bag for 17×17 board.
Anadrome by default uses 7 tiles in rack with 175 tiles in bag for 19×19 board.
Anadrome by default uses 7 tiles in rack with 217 tiles in bag for 21×21 board.
Anadrome by default uses 8 tiles in rack with 362 tiles in bag for 27×27 board.
Anadrome by default uses 10 tiles in rack with 543 tiles in bag for 33×33 board.
Anadrome by default uses 11 tiles in rack with 760 tiles in bag for 39×39 board.
Anadrome by default uses 12 tiles in rack with 1015 tiles in bag for 45×45 board.

Anadrome by default uses 7 tiles in rack with 26 tiles in bag for 7×7 board. There is no other way to change it. Letter distribution, number of tiles in rack and points for each letter are automatically fixed in Anadrome depending upon board size and there is no way to change it manually, even by the board creators. I asked about giving at least board creators the option of customizing all of these things manually by using e-mail, but there is no response from them. At least, I know that they are not inactive, as they had renewed the Anadrome website's https security certificate recently, which had expired on 3rd February 2025.

There are 5 vowels out of 26 tiles in a 7×7 board. There is also half vowel Y, which can be combined with other of consonants in order to make with furthermore shorter words without any of other of vowels too.

26 tiles is minimum number of tiles in bag for any of Anadrome boards.

Even if anyone creates a smaller board than that, may be with any of holes in any of 7×7 boards, as it is possible to design newer boards like that in Anadrome, then, the number of tiles in bag will still be 26 tiles only.

Zitat von Scotty im Beitrag #35
Be careful with this value. The bingo is considered at several other places too, as commented in the code.
By the way, from Configuration → Rules, at least in the latest version 4 of Scrabble3D, it will be actually be only possible to directly change bingo bonus to another fixed integer value, for any of specific games that are played and not to something like 5×Number of tiles in the rack, as it had been in the Anadrome website, isn't it? But, for any of specific games that are played, the bingo bonus will always be a fixed value depending on the number of tiles in the rack whenever that would be compared to the Anadrome website and it will be at least be possible for me to configure that directly, from Configuration → Rules, at least in the latest version 4 of Scrabble3D. But, it will not at least certainly let the computer to automatically take care of that by itself directly in the general case, as it had been in the Anadrome website, for all of general games that are played.

For huge scores, which had been thrilling to play and to acquire in the Anadrome website, the smaller bingo bonus would certainly not make much of a difference.

It would actually certainly make a difference only in lower scoring smaller boards like 15×15 Floor (actual name of board layout had been 15×15 Nullity), 15×15 Zeros (actual name of board layout had been 15×15 Worthless), 21×21 Floor (actual name of board layout had been 21×21 Nullity), 21×21 Zeros (actual name of board layout had been 21×21 Worthless), etc., in which more of connected words and in which more of bingo words would actually only have helped out to score up with higher enough of number of points.


WESPA Player Profile: Raman Viswanathan | Fast Word Study: Google Play Store and Indus App Store | Fast Word Quiz: Google Play Store and Indus App Store

Raman Offline




Beiträge: 291

22.10.2026 19:08
#37 RE: 40 fun bonus tiles as in Anadrome Zitat · Antworten

Zitat von Scotty im Beitrag #33
I think it's a signed integer resp. 2^16. But you need to check the code - and not only the calculation/s but also the feedback.
From user galoomba, in variants channel on Woogles Discord:

Just because Scrabble3D uses signed 16 bit integers and not even signed 32 bit integers to store scores and to index all of possible plays, so that if the total number of possible plays while computing best move exceeds 215-1 = 32767, that will be possible in 3D boards, then an error message is thrown (3D board size is 11×11×11 in the example, that has been shown as above.). Computation of best move and computer opponent players playing their own turns are getting halted after that.

By the way, it looks like signed 16 bit integer has been assigned to the variable used to index all of possible plays. Some of other of variable that has got furthermore number of bits for its own storage has been made use of for displaying the correct total number of possible plays that can be ≥ 215 = 32768 in the Messages tab.


WESPA Player Profile: Raman Viswanathan | Fast Word Study: Google Play Store and Indus App Store | Fast Word Quiz: Google Play Store and Indus App Store

Raman Offline




Beiträge: 291

22.10.2026 19:49
#38 RE: 40 fun bonus tiles as in Anadrome Zitat · Antworten

On 14th April 2026, in variants channel on Woogles Discord:
galoomba: i designed a 3d scrabble board

Thickant: thats cool
Fj00: nobody liked my cube
galoomba: played a bit until the computer decided a blank is too much for it to handle
why is scrabble3d such a bad software
Fj00: no comment
galoomba: oh my god

is the problem that it's using SIXTEEN BIT INTEGERS to index the possible plays
14domino: thats kind of cool
galoomba: what is
probably not the fact that it hangs the whole program if it has 65k available plays
i don't get why it's so slow at generating plays either
i should just go check the code probably
i've seen it list the same play twice...
14domino: cant tell how it works. why doesnt it say flow instead of wolf
galoomba: because af isn't a word?
14domino: no i mean where is the "camera". if its looking from the right it should say wolf but then the letters from actable should be there (E)
galoomba: the slice is shown in the 3d view
it's the 2 plane
just backwards
14domino: ah ok, from the right
galoomba: because the greek letter axis is backwards for some reason
this software is on version 4.0.12 how is it so bad
scrolling is supposed to move the slice but moves 3 rows at a time
engine finally managed to finish a game

11^3 board, 3 standard sets of tiles minus blanks
score 1591-1655
Uruwi: dopinu


WESPA Player Profile: Raman Viswanathan | Fast Word Study: Google Play Store and Indus App Store | Fast Word Quiz: Google Play Store and Indus App Store

Scotty Offline

Administrator


Beiträge: 3.834

23.10.2026 09:06
#39 RE: 40 fun bonus tiles as in Anadrome Zitat · Antworten

Qt's int is 32bit, see https://doc.qt.io/qt-6/qml-int.html I think there is no programming language anymore that defaults to 16bit.


Download: Version 4 | Help:Wiki | Discussion: Forum | News: Mastodon | Fanship: Facebook

Raman Offline




Beiträge: 291

23.10.2026 20:03
#40 RE: 40 fun bonus tiles as in Anadrome Zitat · Antworten


Zitat von Scotty im Beitrag #39
Qt's int is 32bit, see https://doc.qt.io/qt-6/qml-int.html I think there is no programming language anymore that defaults to 16bit.
I just ran some trial runs with high point values for each of letters, although still keeping all of them to be powers of 2, to cover all of possible integer values and with keeping bingo bonus to be 35 points (instead of either 50 points, as it is in conventional Scrabble or 5×Number of tiles in the rack points, as it had been in the Anadrome website) to cover odd integer scores when starting with least score of 2. An alternative to powers of 2 either will be Fibonacci numbers or will be Lucas numbers. Every positive integer can be uniquely represented as sum of distinct powers of 2. Every positive integer can be uniquely represented as sum of distinct non-consecutive Fibonacci numbers (Zeckendorf representation.).

In the latest version 4 of Scrabble3D, feature for re-sizing letters and feature for re-sizing point values for each of letters are yet to be implemented, as it had already existed in earlier version 3.1 of Scrabble3D.

Zitat von Scotty im Beitrag #33
I think it's a signed integer resp. 2^16. But you need to check the code - and not only the calculation/s but also the feedback.
Just because you wrote in post number 33 in this thread that signed 16 bit integer has been used to store the score variable, I had actually mistakenly thought that 215-1 = 32767 is the highest theoretical possible score that Scrabble3D will be able to represent, with the data type that has been assigned to score variable.

I just now checked out that properly at least in latest version 4 of Scrabble3D by using various test runs and that at least latest version 4 of Scrabble3D is properly able to represent all of scores at least till 231-1 = 2147483647 without losing precision, with the data type that has been assigned to score variable.

In the below attached image, the actual scores are 2704481776 for me and 2095139408 for computer. But, just because signed 32 bit integer has been used to store the score variable, after losing precision, my score became -1590485520 and Scrabble3D said computer won this game, whereas actually I won this game.

By the way, I have also tried out a sample 21×21 English Super Scrabble game with 14 tiles in rack, in which more than 215-1 = 32767 total number of possible plays will be possible if one or more blank tiles appear on the rack.

There are no overflows either with the indexing variable of the possible plays or with the variable that shows the total number of possible plays in the Messages tab, at least in the latest version 4 of Scrabble3D, if the total number of possible plays would exceed 215-1 = 32767, with signed 32 bit integer data types that have been assigned to both of those of above mentioned variables.

It looks as if like that the galoomba's error message that I had already mentioned as above actually appeared in the earlier version 3.1 of Scrabble3D, and not actually appeared in the latest version 4 of Scrabble3D. If I remember correctly, then, in the earlier version 3.1 of Scrabble3D, it had been manually impossible to enter in a value of ≥ 1000 for the index of all of possible plays in the toolbar that would have been sorted in the descending order based over upon the scores for each of those of possible plays. As far as I can see, the latest version 4 of Scrabble3D has got none of such of types of issues.

On 14th April 2026, he wrote that in variants channel on Woogles Discord:
Zitat von Raman im Beitrag #38
this software is on version 4.0.12 how is it so bad
So that I had actually mistakenly got confused.

By the way, in the source code of the latest version 4 of Scrabble3D, in what files have the random number generator been used that are used to draw tiles from the bag at random and to order each of players' turns at random?


WESPA Player Profile: Raman Viswanathan | Fast Word Study: Google Play Store and Indus App Store | Fast Word Quiz: Google Play Store and Indus App Store

Scotty Offline

Administrator


Beiträge: 3.834

24.10.2026 09:24
#41 RE: 40 fun bonus tiles as in Anadrome Zitat · Antworten

Zitat von Raman im Beitrag #40
By the way, in the source code of the latest version 4 of Scrabble3D, in what files have the random number generator been used that are used to draw tiles from the bag at random and to order each of players' turns at random?

In ganmeplay.cpp at GamePlay::startNewGame() you find the lines around std::mt19937 rng(nSeed);


Download: Version 4 | Help:Wiki | Discussion: Forum | News: Mastodon | Fanship: Facebook

Raman Offline




Beiträge: 291

24.10.2026 17:42
#42 RE: 40 fun bonus tiles as in Anadrome Zitat · Antworten

Zitat von Scotty im Beitrag #41
In gameplay.cpp at GamePlay::startNewGame() you find the lines around std::mt19937 rng(nSeed);
So, if I understand correctly, then, the random number generation at least in the latest version 4 of Scrabble3D that has been written in Qt6/Qml is actually Mersenne Twister that is actually based over upon the Mersenne Prime number of 219937-1, similar to whatever the random generator that Python makes use of. It produces 53-bit precision floats and it has a period of 219937-1.

So, this, whenever it is being compared to that in the earlier version 3.1 of Scrabble3D that had been written in FreePascal, there had been some of simple mathematical function for the random number generation:
Zitat von Scotty im Beitrag Unfixed bugs in Scrabble3D application.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
 
const
IM = 139968;
IA = 3877;
IC = 29573;
 
function GenRandom(aLimit : longint): longint;
begin
Seed:=(Seed * IA + IC) mod IM;
if aLimit>0 then
Result:=Seed mod aLimit else
Result:=Seed;
end;
 
procedure GenRandomize;
begin
Seed:=GetTickCount mod IM;
end;
 
(that's all)
 



WESPA Player Profile: Raman Viswanathan | Fast Word Study: Google Play Store and Indus App Store | Fast Word Quiz: Google Play Store and Indus App Store

Raman Offline




Beiträge: 291

25.10.2026 11:13
#43 RE: 40 fun bonus tiles as in Anadrome Zitat · Antworten

Zitat von Raman im Beitrag #42
So, if I understand correctly, then, the random number generation at least in the latest version 4 of Scrabble3D that has been written in Qt6/Qml is actually Mersenne Twister that is actually based over upon the Mersenne Prime number of 219937-1, similar to whatever the random generator that Python makes use of. It produces 53-bit precision floats and it has a period of 219937-1.
The advantage of using this random number generator is that for a fixed seed value, it always generates the same sequence of random number values in between 0 (inclusive) and 1 (exclusive). Of course, it will be also be possible to always generate different sequences of random number values every time without fixing a seed value.

Since Python programming language also uses the same random generator, I wonder it would be possible to predict turn orders of each of players in advance and order of drawing of each of letters from the bag in advance by using either Python programming language or Qt6/Qml programming language by itself (at least for debugging purposes).
1. Playing Scrabble with the knowledge of future tiles that will be drawn in Scrabble3D application.
2. Playing Scrabble with no premium squares in Scrabble3D application.
(Name of this of board layout had been 15×15 Fair as in the Anadrome website).
It would be also be certainly fun to play a Scrabble game either against other of human players as opponents or against other of computer players as opponents and try out to maximize the game score, by using the knowledge of the order of the tiles from the bag that will be drawn in advance. Scrabble3D is one of the software applications in which that can be done and at least in the earlier version 3.1 of Scrabble3D that had been written in FreePascal, I had already done that. Although there had been a limit of only 139968 of possible seed values for random number generations, since all of seed values had actually been taken modulo of that of constant value and by using the theory of birthday paradox, collisions in seed values can actually happen at least after 2×√139968 ≈ 748 different games that had been played, there can be certainly be different orders of tiles that can be generated, apart from just changing of those of 139968 of seed values, simply, at least by changing the order of the letters that had been defined in Configuration → Letters.


WESPA Player Profile: Raman Viswanathan | Fast Word Study: Google Play Store and Indus App Store | Fast Word Quiz: Google Play Store and Indus App Store

Raman Offline




Beiträge: 291

25.10.2026 15:03
#44 RE: 40 fun bonus tiles as in Anadrome Zitat · Antworten

Zitat von Raman im Beitrag Math scrabble implemented
It will also help to check out whether Scrabble3D reads all of words in dictionary properly. Alternately, it can also be checked out by Configuration → Word Search, but it is only possible to edit the word there to find out the number corresponding to the word and it is not possible to edit the number there to find out the word corresponding to the number.
Yesterday night, I just realized that the latest version 4 of Scrabble3D does not properly read all of words from loaded dictionary, if they are not in proper sorted order.

If some of words are not properly read from loaded dictionary, then in games that are played against either one or more of computer players as opponents, the computer players do not even play any of those of words on Scrabble board. And either in games that are played against all of human players as opponents or in games that are in human turns that are played against either one or more of computer players as opponents, validation of each of those of words that would be played on Scrabble board would be needed from all of rest of human players whoever are opponents in that game that has been played.

The earlier version 3.1 of Scrabble3D does read all of words from loaded dictionary, although slower, if they are not in proper sorted order and faster, if they are in proper sorted order. The slower reading of words from loaded dictionary, if they are not in proper sorted order would actually make much of difference only if the loaded dictionary contains too much of larger number of words, say in millions, similar to the number of equations in the Maths dictionary that I contributed either with fractional tiles or without fractional tiles.

You can try out the following thing by yourself, at least in the latest version 4 of Scrabble3D. This will certainly let me know whether this issue either occurs only with me or also with you:

Please rename the unsorted.txt file which has been attached along with this post to unsorted.dic and place it in the Scrabble3D dictionaries folder. Load this dictionary into the latest version 4 of Scrabble3D. After that, go to Configuration → Word Search → Search → Word and check out whether MMMCMXCIX (that corresponds to the Roman Numeral of 3999 and that is the last word in unsorted.dic file) and whether XXXVIII (that corresponds to the Roman Numeral of 38 and that is the last word in sorted.dic file) are either in red background colour (invalid word in dictionary, meaning that the word is not properly read whenever loading this dictionary) or in grey background colour (valid word in dictionary, meaning that the word is properly read whenever loading this dictionary).

Please do the same after renaming the sorted.txt file which has been attached along with this post to sorted.dic and after placing it in the Scrabble3D dictionaries folder. If you get red background colour (invalid word in dictionary, meaning that the word is not properly read whenever loading this dictionary) for any of words in dictionary that are not properly read whenever loading this dictionary, then there is some of issue in the latest version 4 of Scrabble3D relating to this that you would need to fix, may be at least during within the next of newer version release of Scrabble3D, some of time later on.

I have just provided to you with a smaller dictionary, although interesting dictionary like Roman Numerals dictionary certainly just for testing purposes. It is also certainly suitable enough for my testing purposes. Although I had actually first diagnosed this issue whenever loading the English dictionary which contained words that are longer than 15 letters in length in unsorted order. The Roman Numerals dictionary may not be suitable enough for publication either into somewhat older SourceForge repository or into somewhat newer GitHub repository, as you had already mentioned some of time 5 years back ago either that you had not been interested in it or that you are afraid of cluttering the configuration file too much.
Zitat von Scotty im Beitrag Math scrabble implemented
Some are interesting but I'm afraid of cluttering the configuration too much.
By the way that I have also included a pre-defined letter set along with both of unsorted.dic file and sorted.dic file certainly just for your testing purposes.

In the latest version 4 of Scrabble3D, it looks as if like some of words like CD (400), D (500), M (1000), MM (2000), MMM (3000) are getting properly read whenever loading unsorted.dic dictionary file, but some of words like I (1), V (5), X (10), L (50), C (100) are not getting properly read whenever loading unsorted.dic dictionary file. It also looks as if like there are none of issues with properly reading whenever loading sorted.dic dictionary file.

All of words that properly exist in unsorted.dic file will be properly be shown in sequential order and that too certainly in grey background colour (valid word in dictionary, meaning that the word is properly read whenever loading this dictionary), whenever traversing sequentially through different numbers in increasing order in Configuration → Word Search → Search → Number, that too cannot be changed manually by typing in a different number in that particular list box field.

Actually only by manually typing in different words in Configuration → Word Search → Search → Word text box field, any of invalid words in dictionary, meaning that the word is not properly read whenever loading this dictionary, will be certainly be shown in red background colour.


WESPA Player Profile: Raman Viswanathan | Fast Word Study: Google Play Store and Indus App Store | Fast Word Quiz: Google Play Store and Indus App Store

Dateianlage:
sorted.txt
unsorted.txt
Raman Offline




Beiträge: 291

25.10.2026 16:16
#45 RE: 40 fun bonus tiles as in Anadrome Zitat · Antworten

Zitat von Raman im Beitrag #38
galoomba: played a bit until the computer decided a blank is too much for it to handle
By the way that in the latest version 4 of Scrabble3D, computer players as opponents playing at different levels, that is playing the moves which score closest to a certain percentage of the best move has already been implemented in Computer Performance in Configuration → Rules.

But that computer players as opponents using the blank tile only if the best move with the blank tile is a certain number of points greater than the best move without the blank tile, is yet to be implemented.


WESPA Player Profile: Raman Viswanathan | Fast Word Study: Google Play Store and Indus App Store | Fast Word Quiz: Google Play Store and Indus App Store

Seiten 1 | 2 | 3 | 4 | 5 | 6
 Sprung  

Wir brauchen Deine Hilfe!

Hallo !

Wir hoffen, dass dir unser Forum gefällt und du dich hier genauso wohlfühlst wie wir.

Um dir und den anderen Mitgliedern das Bestmögliche bieten zu können, würden wir das Forum gern werbefrei gestalten und mit tollen neuen Extras ergänzen.

Wenn du uns dabei unterstützen möchtest, kannst du mit Hilfe einer kleinen Spende dazu beitragen, diese Änderungen zu finanzieren.

Deine Spende hilft!

Spendenziel: 144€
23%
 


Xobor Forum Software von Xobor.de
Einfach ein Forum erstellen
Datenschutz