Thanks for your comments.
The issue with 120 cards limit is a bit more complicated and not easily solvable by throwing computing power at it.
During my days at the university I could prove together with my professor for computational complexity that the problem solved by this program is NP and has a quadratic runtime.
This means that for a problem twice the size it requires 4 times more time and for a problem 3 times the size it requires 9 times more time and so on.
The limit of 120 cards was chosen to be a good compromise between waiting time and result quality. Currently the calculation will always run for at most 5 minutes and return the best result found in this time. For 120 cards most calculations were able to finish within these 5 minutes and return a good result. Beyond that the waiting time would skyrocket and the result quality would decrease, also because some cards may have been sold in the meantime invalidating parts of the calculated suggestion.
As for the leftover cards please send me a message from the website (report problem) to investigate this issue.
I suggest that you order the final suggestion result by seller and check those sellers with only a few cards.
Hi,
I'm the author of the MagicCardWantlist and I've written up some details which also cover your concerns in my blog: http://mcwldev.wente.dk/#post2
This project has been ongoing since 2008 and I've put great efforts into making it safe and easy for everyone to use. If you still have any concerns or questions I will gladly respond.
As for the premium features: This has become necessary to cover the operational (=server) costs of the website. It starts off with only 1,12 euros for one month and it's free for wantlists with less than 10 euros in card value.
Now you can get 1€ worth of credits for only 5 phrases/text-blocks translated. Check out the help pages for more details.
* Import/Export of MCM Wantslists is working again
* Add to shoppingcart function is working again
Both functions required access to the MCM API as they interact directly with a user account. The MagicCardWantlist is now registered as a 3rd Party App and can be granted access for 24 hours. More details follow.
* You can now register with your email address. Your email will also be used to reset your password if you need and for important notifications.
The MCWL account is no longer linked 1:1 to an MCM account. However only one MCM account can be given access through the API.
* Introducing Premium Plans for full access to premium features.
This restores all features like seller names and details for premium users and also enables gives more flexibility. For more details check this page.
* Forced to remove the option to buy premium credits with token cards on MCM.
This was a demand by MagicCardMarket.
* All API keys have been invalidated. You can now grant access to your account (for import/export wants and shopping cart) using the button in your profile for 24 hours.
The functionality is the same as with the API Key before the change, but you can now revoke the access at any time. The access will expire automatically after 24 hours.
The current implementation only uses 3 functions which really use this new authorization option:
1. Import of MCM wants into your MCWL wantlist.
2. Export of your MCWL wantlist into your MCM wants.
3. Add selected offers into your MCM shopping cart.
There may be more features in the future like import/export stock, but there is no schedule yet.
As MagicCardMarket with its new API partially blocked the MagicCardWantlist, the seller's names with a link to their profiles are now show in the shopping suggestions for Premium users.
This is, to make some use out of the calculation, albeit less comfortable.
If MCM's response is positive everything will be reverted as before.
If MCM's response is negative or continues to not respond, the previous model with a flat-rate of 1-2€ for any number of suggestions will be put back.
In this case the shopping will become less comfortable again, as the API access was used to place all suggested items directly in the shopping cart. This was comfortable, fast and reliable as the program itself could check whether all cards made it into the shopping cart. In the previous method you would have to verify the completeness of the order and take the tedious process of identifying the missing items.
Still better than nothing.
Also the import/export of the MCM Wantslists will be removed completely, as it only works with the API.
Please accept me apologies for these inconveniences, but this is something I have little to no influence on.
I think this is the second or third request I've had for full foreign language text, so when I get back from vacation (early December) (and after I fix some MTGImage issues) I will look seriously into adding foreign language text.
Thanks a lot for the suggestions, I'll reply here once I have some updates to share on the foreign text.
I'd like to add one more request for foreign texts and also images.
Keep in mind that the gatherer is horribly incomplete and sometimes even plain wrong when it comes to foreign languages. I've seen printings being labelled as one language but being another, printings for languages that don't really exist (I have yet to see a french unhinged card) and for many older editions no other languages than English are listed (e.g. 4th Ankh of Mishra).
For that matter I've thought about some kind of crowd-sourced card-database where everyone could submit additional data or corrections for review. However if you already thought about this, I would be reinventing the wheel. On the other hand it would be nice to feed this information back.
I've added a pricetrend feature for individual cards to the pricelist. You can search for any card and it will show you the current value, historic values and some general information about the card.
Best price calculator is now freely available for all
Removed All-in-one offers as this functionality is already provided by MCM's own Wantslist and generally not what you want. It also slowed down the site considerably.
When updating prices Premium users now have higher priority than registered users. Registered users have higher priority than anonymous users.
Integrated seller evaluation view
Now you can calculate up to 20 cards for free instead of 5.
Sellers' names for suggested offers are only shown to Premium users for suggestions with a total item value of €10 or more
The view on additional cards from suggested sellers now omits the offers from the current suggestions
Added Journey into Nyx Tokens
Removed a bunch of unnecessary warnings
Fixed a bug in the Java Applet which omitted good sellers under certain circumstances
Greatly reworked the way prices are updated to be much faster. Especially when many users are active you no longer get an error message telling to try again later
Added more server power
Fixed a crash in Java Applet Calculator
--edit: I'm looking for people who are willing to translate parts of the program into their mother tongue in exchange for some credits.
Direct and fast import from and export to MCM Wants
The function to add suggestions to the MCM shopping cart is now much faster and verifies that all articles have been added
Every registered user now has a private profile page. On this you can review your premium transactions and save the MCM API Key which is needed for both above functions.
There's now an alternative calculation method using a Java Applet. For small lists it creates much cheaper results. Being a Brute-Force approach it doesn't work well for medium to large lists though.
The issue with 120 cards limit is a bit more complicated and not easily solvable by throwing computing power at it.
During my days at the university I could prove together with my professor for computational complexity that the problem solved by this program is NP and has a quadratic runtime.
This means that for a problem twice the size it requires 4 times more time and for a problem 3 times the size it requires 9 times more time and so on.
The limit of 120 cards was chosen to be a good compromise between waiting time and result quality. Currently the calculation will always run for at most 5 minutes and return the best result found in this time. For 120 cards most calculations were able to finish within these 5 minutes and return a good result. Beyond that the waiting time would skyrocket and the result quality would decrease, also because some cards may have been sold in the meantime invalidating parts of the calculated suggestion.
As for the leftover cards please send me a message from the website (report problem) to investigate this issue.
I suggest that you order the final suggestion result by seller and check those sellers with only a few cards.
I'm the author of the MagicCardWantlist and I've written up some details which also cover your concerns in my blog: http://mcwldev.wente.dk/#post2
This project has been ongoing since 2008 and I've put great efforts into making it safe and easy for everyone to use. If you still have any concerns or questions I will gladly respond.
As for the premium features: This has become necessary to cover the operational (=server) costs of the website. It starts off with only 1,12 euros for one month and it's free for wantlists with less than 10 euros in card value.
I've also fixed an issue regarding import/export of MCM wants lists.
Along a bunch of changes were necessary:
This is, to make some use out of the calculation, albeit less comfortable.
If MCM's response is positive everything will be reverted as before.
If MCM's response is negative or continues to not respond, the previous model with a flat-rate of 1-2€ for any number of suggestions will be put back.
In this case the shopping will become less comfortable again, as the API access was used to place all suggested items directly in the shopping cart. This was comfortable, fast and reliable as the program itself could check whether all cards made it into the shopping cart. In the previous method you would have to verify the completeness of the order and take the tedious process of identifying the missing items.
Still better than nothing.
Also the import/export of the MCM Wantslists will be removed completely, as it only works with the API.
Please accept me apologies for these inconveniences, but this is something I have little to no influence on.
I'd like to add one more request for foreign texts and also images.
Keep in mind that the gatherer is horribly incomplete and sometimes even plain wrong when it comes to foreign languages. I've seen printings being labelled as one language but being another, printings for languages that don't really exist (I have yet to see a french unhinged card) and for many older editions no other languages than English are listed (e.g. 4th Ankh of Mishra).
For that matter I've thought about some kind of crowd-sourced card-database where everyone could submit additional data or corrections for review. However if you already thought about this, I would be reinventing the wheel. On the other hand it would be nice to feed this information back.
Example: Thundermaw Hellkite
If desired I can make a trimmed down version to embed in other websites.
I'm still looking for translators.
--edit: I'm looking for people who are willing to translate parts of the program into their mother tongue in exchange for some credits.
I've added a Translation Credits option, where you can gain credits by submitting translations.
A new version of the MagicCardWantlist has been published
The most important Changes are:
Feedback is always welcome.