Forge Beta: 06-24-2011 ver 1.0.14 SVN rev 10003

[Chris H.]

I can't even run this version. Whenever I double click the forge.exe launcher, I get a java heap space error. I haven't had any problems whatsoever with heap space errors on the computer I'm using now until this version.

`
We are sorry to hear that you are having problems Kersed. Can you try this and let us know if it works:

Rename the jar file to "run-forge.jar"

Move the "forge.exe" launcher from the previous version of forge into the folder containing the new version of forge. This should overwrite the launcher that came with the newest version of forge.

You should now be able to launch the renamed jar file from the newest version using the launcher that came with the earlier version of forge.

If this does not work you may want to re-download and extract the archive into a new folder.

[Braids]

@Chris H.: i think this is a very important bug. thank you for taking the time to help Kersed out.

is there a way to look inside the jar and/or exe files created with this beta, to see if there are any unnecessary files in there? i saw some java source files in my bin directory earlier today. i still don't know if that was my error or not. Rob Cashwalker did say that the build was unusually large. this may be a big clue.

by the way, i found several places to improve performance and save heap space today. i'm not sure if they'll help with startup. the specifics: there are lots of calls to CardFactory.getAllCards, which makes a shallow copy of the master list. truth is, most code doesn't modify the result, so the copy is unnecessary. they could call CardFactory.getCards instead. i flagged the calls i thought could be changed with TODO comments in my upcoming lazy CardFactory update.

[Sloth]

by the way, i found several places to improve performance and save heap space today. i'm not sure if they'll help with startup. the specifics: there are lots of calls to CardFactory.getAllCards, which makes a shallow copy of the master list. truth is, most code doesn't modify the result, so the copy is unnecessary. they could call CardFactory.getCards instead. i flagged the calls i thought could be changed with TODO comments in my upcoming lazy CardFactory update.

I would be very grateful for every performance boost/heap space reduction, since my system is slowly getting too slow for the game.

[Chris H.]

@Chris H.: i think this is a very important bug. thank you for taking the time to help Kersed out.

is there a way to look inside the jar and/or exe files created with this beta, to see if there are any unnecessary files in there? i saw some java source files in my bin directory earlier today. i still don't know if that was my error or not. Rob Cashwalker did say that the build was unusually large. this may be a big clue.

by the way, i found several places to improve performance and save heap space today. i'm not sure if they'll help with startup. the specifics: there are lots of calls to CardFactory.getAllCards, which makes a shallow copy of the master list. truth is, most code doesn't modify the result, so the copy is unnecessary. they could call CardFactory.getCards instead. i flagged the calls i thought could be changed with TODO comments in my upcoming lazy CardFactory update.

`
Thank you.

I see that the forge-1.0.14.tar.bz2 file has been downloaded 132 times. So far Kersed is the only person that has reported having a problem launching forge and getting a java heap space error.

I hope that he can get this straightened out and enjoy the new version.

Oh, will the CardFactory update by chance also save some heap space?

[Braids]

Oh, will the CardFactory update by chance also save some heap space?

progressively. i wish i had better news.

the problem is that code that edits decks, generates booster packs, drafts boosters, downloads decks (?), generates decks of certain colors, and the Bazaar of Baghdad (the card shop) all operate on the entire set of all cards.

i do know other ways to Twincast less memory. we could use programming models like generators (a la Python prog lang) for filtering lists of cards without creating any unnecessary list Clone s. i see many many occasions of copying CardList instances when filtering. that model only works when we need to iterate over the filtered list, though. we can't sort it, which means the deck editor is still a Llanowar Behemoth.

[Kersed]

I can't even run this version. Whenever I double click the forge.exe launcher, I get a java heap space error. I haven't had any problems whatsoever with heap space errors on the computer I'm using now until this version.

`
We are sorry to hear that you are having problems Kersed. Can you try this and let us know if it works:

Rename the jar file to "run-forge.jar"

Awesome. This worked. Thanks a ton. I'm not sure why I didn't try this earlier.

[Chris H.]

Awesome. This worked. Thanks a ton. I'm not sure why I didn't try this earlier.

`
Great, now play some forge and enjoy this new version.

[Rob Cashwalker]

Chris, I'm not even bothering to repackage this release until the forge.exe thing is straightened out. Based on the filesize, it really seems like the jar has been compiled into the exe. How it's related to any difference in performance, I don't know.

Braids- the getAllCards thing in CardFactory ALWAYS bothered me.... The deck editors don't NEED fully functional card objects, they just needs stats and text. The deck generators and booster generators just need the basic properties.
There are 2 uses I have for functional card objects outside of the game - in the draft and sealed AI - keywords and abilities attached to the card are counted in a card evaluation and mana abilities of lands are checked to see if they fit the colors of the deck.
Anything you can do along these lines is more than welcomed, whether it directly relates to performance or not.

The biggest memory hog is the ImageCache, put a breakpoint in the routine which puts mana symbols onto cards, and you see that despite the concept of "cache", the images are recreated anytime one detail in the game is changed....

[jendave]

Chris, I'm not even bothering to repackage this release until the forge.exe thing is straightened out. Based on the filesize, it really seems like the jar has been compiled into the exe. How it's related to any difference in performance, I don't know.

Strange. This is a problem. I'll fix it.

[jendave]

Chris, I'm not even bothering to repackage this release until the forge.exe thing is straightened out. Based on the filesize, it really seems like the jar has been compiled into the exe. How it's related to any difference in performance, I don't know.

Strange. This is a problem. I'll fix it.

Problem fixed. There was a bug in the launch4j maven plugin. I fixed that so now the Windows exe no longer contains the jar file. I also removed unnecessary dependencies from the final jar file. This makes the archive files about 50% smaller. I tested the Windows exe on Windows XP with 4GB of RAM. It ran fine.

[Rob Cashwalker]

thanks.

Chris... new package?

[Chris H.]

thanks.

Chris... new package?

`
I will go ahead and release a fresh new beta as a test.

You got the draft pick data collection added and we can get people started.

[Rob Cashwalker]

Yeah.... I'm really curious how this goes...

[Chris H.]

Yeah.... I'm really curious how this goes...

`
Yeah, we all are.

Having the basic hinting system helps, adding a min-max system can improve things.

But having some real world rankings and ratings is also helpful.

[Chris H.]

Braids is doing a major merge starting at rev 10108. I think that I should wait until this merge is finished as there is a note that "Still need to fix compile".

Let me give Braids some more time. Braids will see this message and will let us know that this work is finished.

I have a dentist appointment in a couple of hours ... get a tooth ground down for a post and crown.

Sometime this evening the rate of merges will slow down and I will be able to build and upload of the beta using Dave's fixed Maven script.