Poker Hand XML

[LooselyBased]

[ QUOTE ]

I noticed we are using "Ah" instead of Card "A" Suit "h". Just wanted to make sure this was the way we wanted to go. I think it's fine but I know some good points were made to split it.


[/ QUOTE ]

The short answer would be, what suit is a Joker?

If you assume that this is in fact a general poker HH format, then you must consider jokers as they are part of some decks/games.


Also, I still have not seen an official schema listed anywhere outside this thread. Who is maintaining it?

Card "A" Suit "h" - forgive me, I know not XML - why would a concatenated version be in any way preferable to separate fields?

[pothead]

Great initiative!
I'm working on a replayer and will for sure try to go with this format for input.

I just have skimmed through the thread a bit so far, but a couple of comments:

"Ah" is compact and does the job...I don't see that it would be much easier to use if it was separate attributes.
And if one like to add games with jokers I guess either way would work if the suit is optional, but it would make more sense to use the concatenated form for sure.

Maybe one way of incorporating dealer chat, reads and comments would be something along these lines:
<font class="small">Code:</font><hr /><pre>
&lt;actions&gt;
&lt;action type="bet" id="player1" amount="2900" allin="true" /&gt;
&lt;meta type="chat" id="player1"&gt;Call me plz!&lt;/meta&gt;
&lt;meta type="read" id="player2"&gt;Gives up his blinds easily&lt;/meta&gt;
&lt;meta type="comment"&gt;Of course I'm hoping for a fold here&lt;/meta&gt;
&lt;meta type="dealer"&gt;Player2 is disconnected and is considered all-in&lt;/meta&gt;
&lt;meta type="observer" id="AAAAllIn"&gt;Give me money PLZ!!!&lt;/meta&gt;
&lt;/actions&gt;
</pre><hr />
For example table reads or player stats as well as any chat happening before the hand starts could be added in the preflop round for most purposes I guess, but maybe it would be nice to be able to have meta elements also in the results?

I will read through the thread a bit more and start putting together a RelaxNG schema in the next few days since I need to have something to work with ASAP.
Would be great though if I could get it as close a possible to the final PokerHandXML spec, so I hope there will be a consensus soon.
It seems pretty much there and it would be great if it could be wrapped up into a proper schema.

[LooselyBased]

After looking over the most recent schema, I have a few issues with the general approach that's being taken.

You should not be using strings to represent everything, especially when there is a more strongly-typed way of doing this (e.g., rather than making the "allin" field a string, it should be a boolean). The reasoning here is two fold:

1) Web services middleware and databases have benefited greatly from research into how to compress XML files. Part of this compression relies on converting the files to binary format which minimizes overhead. To continue with my initial example, if you have a string ("true" or "false"), you have to store 4 or 5 characters. However, with a boolean you can theoretically store just one bit (though in practice it's usually a byte).

2) Some code generators (e.g., Castor for Java, Visual Studio for C#) automatically generate code based upon an XML schema. By not providing strongly-typed fields, you're essentially forcing the programmer to write wrapper code for the generated data structure. Consider the following code which might be generated by Castor:

<font class="small">Code:</font><hr /><pre>
public class Blind {
...
String getAllIn() { return _allin; }
}
</pre><hr />

Now the programmer is forced to either do a string-comparison at runtime (inefficient) or build his/her own wrapper data structure (e.g., BlindWrapper) which takes a Blind object in its constructor and does the conversion once. This approach is also inefficient as it causes lots of duplicated resources.


Since poker hand databases are likely to be significant in size, it's extremely important that both compression and code generation are efficient.

[mikechops]

[ QUOTE ]
it will be a web based converter, just as legopoker is set up now.

as for setting up a web service - not sure if I will do this.

No it will not be open source.

[/ QUOTE ]

That's a bit of a disappointment on both counts. I guess web-based is fine for some apps but I wouldn't be able to make the Realtime HUD work.

Anybody interested in collaborating on a .dll based open source version?

[LooselyBased]

[ QUOTE ]

Anybody interested in collaborating on a .dll based open source version?

[/ QUOTE ]

Sure. I had already planned on doing this, but it'd be great to have someone to collaborate with.

[tangerine]

i didnt really follow all posts on this threads and HM threads so.....

[ QUOTE ]
it will be a web based converter, just as legopoker is set up now.

[/ QUOTE ]
so it will be converting most online poker rooms' HH into Poker Hand XML standard as discussed?
[ QUOTE ]

That's a bit of a disappointment on both counts. I guess web-based is fine for some apps but I wouldn't be able to make the Realtime HUD work.

Anybody interested in collaborating on a .dll based open source version?

[/ QUOTE ]
how would this work? so we expect the converter to save all the pokerhand xml that it has converted? and allows our web app to access them? sounds cool~ i would be happy to contribute if i can

[LooselyBased]

[ QUOTE ]

how would this work? so we expect the converter to save all the pokerhand xml that it has converted? and allows our web app to access them? sounds cool~ i would be happy to contribute if i can

[/ QUOTE ]

The idea behind web services is to allow remote access to proprietary services. In the context of the converter, you would be able to access a remote method which takes a site-specific hand history (e.g., Party Poker, Full Tilt, etc.) and it would return you the appropriate XML format. Potentially, a bi-directional converter could be offered as well.

On the other hand, since it's an open format and we're trying to facilitate widespread adoption, it's likely better to have an open library. Personally, I prefer C# for my development platform, so I'll be implementing a converter library in that. However, the bulk of the code can easily be generated automatically for other languages.

[LooselyBased]

Quick question: what is the "format" field supposed to represent? It looks like it's cash game/MTT/SNG, is that right?

[Tickner]

from experience (im working on LP converter now), creating a converter for every site and every hand hsitory is MANY HOURS of hard work. To those offering to create an open source converter - thank you.

That said, my goal of creating this thread was simply to establish a format - I have never committed to create an open source web service as some here seem to think I have.

I'm quite busy this week but next week I will put more effort into the schema etc of this. If someone else wants to take the lead and create one that'd be cool too.

[Tickner]

[ QUOTE ]
i didnt really follow all posts on this threads and HM threads so.....

[ QUOTE ]
it will be a web based converter, just as legopoker is set up now.

[/ QUOTE ]

so it will be converting most online poker rooms' HH into Poker Hand XML standard as discussed?


[/ QUOTE ]

yes, thru the web interface, in much the same manner as it converts the hand histories into 2p2 format or HTML format.