Cockatrice - intended as successor to MWS

[AmbivalentDuck]

2) The fact that you can arrange your lands how you like without that information being contained in the protocol was considered important during early development. I think we're both saying the same thing, however, there may not be enough arrangement modes at the moment so that the one you want is not there. Essentially, for everybody, there has to be a function which maps the "land index" (1,2,3) to a point on the table (and backwards). The purpose of this is not only that your opponent sees *your* lands how *he* likes (that is only a side effect), but that you can double-click on cards in your hand, placing them on the table at the right point (which is determined by the aforementioned index) with regards to your arrangement preferences. That being said, how would you like your lands to be placed? If you want just the horizontal arrangement like the other table rows have it, you can already have that by unsetting that checkbox in the settings dialog. I'm sorry if I misunderstood what you said.

Yeah, I probably just need to contribute a layout.

3) If you have card pictures installed, you can already do that by changing the order of the sets (there is a separate window for that accessible through the deck editor). Maybe it can be extended so that you can sort the sets for each card individually.

Individual sorts would be nice...and even beyond that...art version.  Ie.  I want to play with my preferred Island art.

4) I think this could be a VERY difficult task. However, it would technically be possible to implement that via the existing client code. The protocol code (but only the new one, which is not ready for release yet) is encapsulated in classes, which can easily be used by another program. Decision harvesting could be done by joining a game as a spectator and gathering all the events. Of course, you need to be able to look at one player's hand to do that (which spectators cannot do at the moment for obvious reasons), but I don't regard that a major concern.

I agree: the biggest issue is you need to know what's in at least one hand, the rest is just writing a custom client.  Can hand information be accessed *after* the game is completed via server log or some such?

[mbruker]

Can hand information be accessed *after* the game is completed via server log or some such?

Yes. I'm probably going to record the whole game on the server with all information open. I want to be able to look at a replay of my games anyway, so why not save it on the server. That replay will be accessible in the same data format as the in-game protocol stream and thus readable by the existing client classes.

[mbruker]

Just so you don't think there's nothing going on: The new protocol implementation is finished for the most part, but it's still going to take some days until it's fully usable in the client. BTW, sorry for screwing up the git master branch.
Preview: http://brukie.de/cockatrice/Bildschirmfoto16.png
(Edit: updated screenshot)

[Shean]

Given the fact that I play on MWS all the time and hate it, I'm very excited about this project.

I see that you're having people compile the code themselves.  I'm totally noob to this, so I hope a precompiled version comes out sometime.

I really hope you fixed the shuffler!

[mbruker]

The Cockatrice server is going offline for a couple of hours (this includes the web server). After that, an additional experimental server using the new protocol code will be available at port 4748.
The current git version of the client is usable, but there are some minor problems and unfinished pieces that still need to be fixed. I hope we can release a new stable version of the client soon.

I hope a precompiled version comes out sometime.

For Windows users, there is a precompiled version available, though it's not always very up-to-date since we're changing a lot of small things and don't see a reason to have the users update all the time.
A precompiled version for Mac would also be interesting, however, we don't have access to a Mac to do that. It would be appreciated if someone (with Mac packaging experience) volunteered to do that for us.
A precompiled version for other operating systems doesn't make sense since that should be left for the respective distributor to do. Else, one would face a lot of compatibility problems (libraries, paths etc).

I really hope you fixed the shuffler!

What's there to fix? I've never run any statistical tests on the MWS shuffler, so I don't know whether it works correctly. I don't know that about the Cockatrice shuffler, either. That is a very minor concern though and will be checked once all the features are in place.

[Shean]

For Windows users, there is a precompiled version available, though it's not always very up-to-date since we're changing a lot of small things and don't see a reason to have the users update all the time.

Yes, I found this and was able to make it work.  The documentation provided isn't very user-friendly, but after figuring out that I needed to run the Oracle.exe program to get the cardbase, it works.

What's there to fix? I've never run any statistical tests on the MWS shuffler, so I don't know whether it works correctly. I don't know that about the Cockatrice shuffler, either. That is a very minor concern though and will be checked once all the features are in place.

As for "statistical tests," I haven't run any either.  However, I can tell you that the MWS shuffler is terrible.  I mulligan 200%-300% more on MWS than I do with real cards.  I've also run into having 4 of the same card in my hand multiple times on MWS, which I have never seen with real cards.

[Godder]

In my experience with Apprentice, MWS and so on, their shufflers are excellent - it's our shuffling that's not up to speed.

[miseiler]

In my experience with Apprentice, MWS and so on, their shufflers are excellent - it's our shuffling that's not up to speed.

This is, sadly, the truth. You simply don't get true randomness from riffle shuffling, except after ~9 (!) shuffles in a 60-card deck: http://www.jstor.org/pss/2665604

Your deck will simply behave differently online and off because of this.

[noisome]

In my experience with Apprentice, MWS and so on, their shufflers are excellent - it's our shuffling that's not up to speed.

This is, sadly, the truth. You simply don't get true randomness from riffle shuffling, except after ~9 (!) shuffles in a 60-card deck: http://www.jstor.org/pss/2665604

Your deck will simply behave differently online and off because of this.

I haven't found the MWS shuffler to be excellent.  I would prefer to see the code however before passing true judgment on it.  Now if a program can effectively riffle shuffle several times along with overhand shuffle with random card counting during shuffling, I believe this will maximize the shuffling randomness.

Also, it may take ~9 riffle shuffles for randomness flipping only one card down at a time per side.  It doesn't account for multiple cards down at one time and the same for the other side. Also it does not include overhand shuffling.  Many magic players also do pile shuffling.  How about incorporate all of these in a shuffler?  I believe my shuffles IRL are slightly better than MWS.

[mbruker]

Now if a program can effectively riffle shuffle several times along with overhand shuffle with random card counting during shuffling, I believe this will maximize the shuffling randomness.

You seem to misunderstand how random shuffling works.
All programmatic shufflers behave in the same way (with the exception of effects that come from less than optimal RNGs, but I'm pretty sure that is not a concern here). They do not try to imitate human shuffling patterns; on the contrary, human players need these shuffling patterns to imitate what these shufflers do, which is to put each card in a random place in your deck regardless of the positions of the other cards. It should be clear that this goal is not achievable by just pushing one half of your deck into the other several times, though it comes close enough for us to not notice.
Let me explain how the common approach works. It would (yet only in principle) be possible to achieve random shuffling with real cards. To do that, you need a die with 60 or more sides and a uniform probability distribution (doesn't exist, but you can simulate one pretty well, though I'm not going to elaborate on that now). Roll the die until you have a number "x" that is less than or equal to the number of cards left in your deck. Take the x-th card out of your deck and put it on top of the other pile. Repeat everything until your deck is empty, then make the other pile your shuffled deck.
Though the method obviously depends on the initial order of cards in the deck, it is not necessary to do any preparation or additional shuffling as long as the quality of the random number generator is sufficient.
By the way, there are very obvious and easy methods to eliminate first order deviations from a uniform distribution in a given list of random numbers. If you compare two of the numbers, you get a uniformly distributed number that has one of two possible values (first number higher than second and vice versa). Repeat this as often as you need to increase the number of possible values. This is why you should never let Mana Crypt deal damage to you when you roll an odd (or even) number, but you should roll twice with the same die and compare the numbers. I have seen dice that could only roll fours and sixes; using these dice, Mana Crypt would not love you (or maybe it would).

Seeing that these statistics seem to be a major concern for many of you (understandably), I will do a statistical analysis on different shufflers and maybe write a more detailed article about that.

[noisome]

Actually, I understand exactly how random shuffling works, and despite the effectiveness of a RNG, each program and machine's quality is different.  

A lot of shufflers effectively create an array/list of each card in the deck and pop a random number out of the length of the array and place into a temporary array/list to create a randomized deck.  They could even random sort the main array.  And sure, this works.  There is no reason for it not to work.

However good this seems to work, it doesn't work for MWS, however they do it.  It may work on some machines, but on mine it doesn't seem to work well.  It also doesn't seem to work well on a lot of others according to the different forum posts.

Its whatever you want to do, I'm not creating the program.  My suggestions of shuffling are of opinion only and have no science behind them other than how many times a human needs to shuffle to make apparently "true" randomization.  If humans can achieve this, then why not mimick it to gain full effectiveness?

Statistical analysis would be great if you get around to it, though.

[AmbivalentDuck]

If you wanted truly random, you could use the 6th digit of the standard deviation of the CPU temp over  a ten minute interval or some such.

[mbruker]

despite the effectiveness of a RNG, each [...] machine's quality is different.

No. The output of an RNG and subsequent calculations are deterministic.
It doesn't matter anyway since without an analysis of either the exact code being used or its results, this discussion will not lead anywhere. I will work on that topic once the next stable version of Cockatrice is ready; I'm afraid there are still a few minor bugs left.
I would also like to note that if the shuffling code behaves in a less than optimal way, at least both players have to suffer from it since it is done in the server. However, this might be different in MWS.

[P]

Is there any Network support for opening/saving decks? The user interface seems a bit rudimentary, lacks keyboard shortcut support (customs would be hot!), but the playing itself seems fine so far.

Good job!

[mbruker]

The new version with the reworked protocol code is now ready. There is a precompiled version available on the download page. I apologize for the enormous delay.
 - Sideboarding is not implemented yet, however, the screen where it will be done is in place, and the rest is a matter of days as soon as there is some time.
 - Storing and loading decks on the server works. To be able to use the deck storage tab, you need to be logged in with a valid user name and password. You can use the same data that you do on this website.
 - Arrows work. You can create an arrow by right-clicking a card and dragging onto the target card. Pressing a modifier key (shift, alt, ctrl) before doing so changes the color of the arrow.
 - The user interface is now more abstract so that you can be in multiple games and chat channels at once.
 - If you are logged in with a valid user name, you can now disconnect from the server and still be in your games after you reconnect. This works, but still needs minor improvement.
This change log is far from complete, but you'll notice most changes yourselves anyway. Keep in mind that this version (like all the ones before) is considered experimental and not everything has been tested thoroughly.
 
The server that is suitable for this version is now running on port 4747 again.
I know that not all suggestions have been implemented yet, but most will be eventually. If you are missing a feature that has already been suggested, please be patient

Is there any Network support for opening/saving decks?

Sorry, I don't understand the question.

The user interface [...] lacks keyboard shortcut support

In fact, most actions do have a keyboard shortcut, you just can't change what it is. This will eventually be made configurable.

[Yare]

Glad to see that progress is being made on this. Thank you for your work!

[mbruker]

Sorry for not posting for such a long time, but I've been pretty busy. Development is going to continue now, and I got a whole screen full of issues that need to be addressed.
I know it's not exactly what you've been asking for, but I had fun with arrows yesterday: http://brukie.de/cockatrice/Bildschirmfoto22.png

[PhilipJFry]

I feel un-smart - I downloaded this last night, but cannot get any cards to show up in the deck editor.  I am sure I did something wrong, but I don't know what.  Please help me - I am un-smart.

[CorwinB]

I feel un-smart - I downloaded this last night, but cannot get any cards to show up in the deck editor.  I am sure I did something wrong, but I don't know what.  Please help me - I am un-smart.

You have to import the cards running the "Oracle.exe" program.

[PhilipJFry]

I feel un-smart - I downloaded this last night, but cannot get any cards to show up in the deck editor.  I am sure I did something wrong, but I don't know what.  Please help me - I am un-smart.

You have to import the cards running the "Oracle.exe" program.

I think I did that.  I found the Oracle.exe program and ran it, it retrieved data from all sorts of Magic sets.  But when I ran Cockatrice.exe, and opened the deck editor, there were no cards in it.  I typed titles into the search function and hit enter, but nothing showed up. 

[CorwinB]

I think I did that.  I found the Oracle.exe program and ran it, it retrieved data from all sorts of Magic sets.  But when I ran Cockatrice.exe, and opened the deck editor, there were no cards in it.  I typed titles into the search function and hit enter, but nothing showed up. 

What is the size of the cards.xml file in the "oracle" folder, after running oracle.exe ? Perhaps the program is blocked by your firewall from getting the data ?

[mbruker]

I would guess that you simply didn't set the path to the cards.xml file correctly. You need to do this in the configuration dialog after generating the file. The same is true for the pics directory.

[PhilipJFry]

I would guess that you simply didn't set the path to the cards.xml file correctly. You need to do this in the configuration dialog after generating the file. The same is true for the pics directory.

Considering I have no idea what that means, I am going to guess this is the culprit.  Can you tell me how to do this?

[CorwinB]

I would guess that you simply didn't set the path to the cards.xml file correctly. You need to do this in the configuration dialog after generating the file. The same is true for the pics directory.

Considering I have no idea what that means, I am going to guess this is the culprit.  Can you tell me how to do this?

Run the program, go in the "Cockatrice" menu, then "Settings". There are 3 fields for Deck directory, Pictures Directory, and Card Database. Use the "Card Database" field to point it to the "cards.xml" file in your oracle folder. Don't forget the "Download cards picture on the fly" checkbox.

[PhilipJFry]

I would guess that you simply didn't set the path to the cards.xml file correctly. You need to do this in the configuration dialog after generating the file. The same is true for the pics directory.

Considering I have no idea what that means, I am going to guess this is the culprit.  Can you tell me how to do this?

Run the program, go in the "Cockatrice" menu, then "Settings". There are 3 fields for Deck directory, Pictures Directory, and Card Database. Use the "Card Database" field to point it to the "cards.xml" file in your oracle folder. Don't forget the "Download cards picture on the fly" checkbox.

That was the problem, the fix worked great, and now I am in.  Thanks!

[mbruker]

We are getting closer to releasing a new version.
- Sideboarding works.
- The zone view code is undergoing heavy changes so that the table doesn't have to shift any more when you open a view, all the while remaining clearly arranged.
- Lots of minor improvements and fixes.

Screenshot: http://brukie.de/cockatrice/Bildschirmfoto26.png

[CorwinB]

We are getting closer to releasing a new version.
- Sideboarding works.
- The zone view code is undergoing heavy changes so that the table doesn't have to shift any more when you open a view, all the while remaining clearly arranged.
- Lots of minor improvements and fixes.

Screenshot: http://brukie.de/cockatrice/Bildschirmfoto26.png

That looks great ! I use dual instances of Cockatrice for solo testing (easier than proxying or sleeving up cards), and I'm very excited to see sideboarding implemented.

Regards,

[mbruker]

There is a new version on the download page. I apologize for the delay.
Apart from sideboarding, which is fully working now, there have been a lot of minor improvements and changes. Be sure to run the oracle tool after updating so that you can make use of some new card properties.
The Cockatrice server has been updated as well, so you will need this version to connect to it.
Linux and Mac users, please compile the code yourselves as usual if your distribution does not contain a precompiled package (which I believe no distribution does at the moment).

[JDawg13]

Just downloaded this and it's really cool!  Thanks for taking the time to put it together.  The on-the-fly card art downloading is awesome, so much easier than having to manually download pictures for each new set that comes out for MWS.  Haven't played any games with it yet, but I'm impressed with what I've seen so far.

EDIT:  One thing, when I go to play a game, the black space takes up a lot of the screen making it kind of hard to see which card is which without mousing over it to get the zoomed in view.  Here's what it looks like for me: http://i41.tinypic.com/1zxbfro.jpg.  Is there any way to reduce the black space?  Thanks!

[mbruker]

The black space you're referring to is needed for your opponent's part of the table. The cards won't get any bigger, I'm afraid. You can get a better picture quality though by manually downloading HQ pictures.