Page 1 of 2 1 2 LastLast
Results 1 to 10 of 19

Thread: ETA On Web API?

  1. #1
    Basic Member
    Join Date
    Sep 2012
    Posts
    10

    ETA On Web API?

    Is there any estimate when this will be back? Any rough guess would be helpful for us. Is it going to be this year?
    Last edited by googlexx; 10-05-2012 at 04:22 AM.

  2. #2
    Basic Member
    Join Date
    Nov 2011
    Posts
    165
    This api cannot be that hard to get up. Hire an intern...

  3. #3
    Basic Member
    Join Date
    Nov 2011
    Posts
    149
    The difficulty is not in turning it on, but in coping with the demand. Hammering the API was causing in game performance to suffer (at a guess game history pages etc) and working out how to deal with that is going to take time to think about, to fix and potentially additional infrastructure. At a guess the API is on the back burner while they do the game/TI2 documentary/tutorial stuff/the next custom event map/more trailers/etc.

  4. #4
    Basic Member
    Join Date
    Nov 2011
    Posts
    165
    Again, this is not a hard problem to solve. In the meantime, it's unfair they allow dotabuff to scrape their data but other not other developers.

  5. #5
    Basic Member Shinkai's Avatar
    Join Date
    May 2012
    Posts
    186
    They do not allow dotabuff any more than anyone else, [they,dotabuff] just use different mechanism to do so. It was mentioned multiple times on the forums earlier

  6. #6
    Basic Member VeRychard's Avatar
    Join Date
    Feb 2012
    Posts
    292
    I'd like to know that different mechanism, because it would surely speed up some processes
    There are more than 1 sites like that which uses the same(?) mechanism, so it must be something easy
    >Click< here to make a clicking sound.

    Overall average (K+A)/D and K/D tracker, and a MMR solution?


  7. #7
    Basic Member Shinkai's Avatar
    Join Date
    May 2012
    Posts
    186
    http://dev.dota2.com/showthread.php?...l=1#post363391 I think they use this one, all that stuff should be mentioned in that thread, have fun
    Last edited by Shinkai; 11-23-2012 at 04:13 AM. Reason: better link, wording

  8. #8
    Basic Member
    Join Date
    Nov 2011
    Posts
    165
    Quote Originally Posted by Shinkai View Post
    http://dev.dota2.com/showthread.php?...l=1#post363391 I think they use this one, all that stuff should be mentioned in that thread, have fun
    Thanks!

  9. #9
    Basic Member
    Join Date
    Nov 2011
    Posts
    165
    It seems like this requires Steam to be running on the server, so it's not really optimal. It would probably require a windows server, although I'm not positive - Steam runs on linux but I don't know if I can mimic the Dota requests, since you can't install Dota 2 on linux. I'm not really too familiar with this method though. I'd then need two servers, one which runs Steam and has a C# service that replicates the web service API, caches, etc, and then a second linux python/ruby/node/whatever server to query it and run the web server. So much work for what essentially will be throwaway if the API is ever released.

    A better method would be to just scrape dotabuff, which is kind of evil but the only option at the moment. There is still missing data though, such as match skill level.

    As an aside, the API needs to be updated to handle skill level properly. You can search by skill level, but matches don't return the skill level of the games. Therefore the consumer has to call the same query n times, where n is the number of skill levels. It seems like Valve can't make up it's mind about whether or not people should see your skill level. Make up your mind and remove match skill level from the query fields, or include it in the results : )

  10. #10
    Basic Member
    Join Date
    Nov 2012
    Posts
    35
    Scraping dotabuff is indeed bad!
    However, today is your lucky day - in the next few days we will be releasing a data dump of the entire match details from game number 1 to the most recent (somewhere around 70,000,000).

    Pre-emptive FAQ:

    What does the data contain?
    It's nearly an exact copy of the data you can get from the API for match details for now, with a few minor additions

    How big is the data?
    Big. The entire dataset is between 50Gb and 100Gb, and growing at ~500,000 games a day which is about a 1Gb/day.

    How will the data be shared?
    We're working on that exact problem now. Most likely is some form of P2P sharing for bulk downloading (updated with weekly dumps) and then some method of getting only the most recent games.

    How are you getting the data?
    We can't say. According to CyborgMatt, Valve are OK for us to scrape and share, but not to let people know how we're doing it. Obviously this shouldn't be an issue since we're making the data available to everyone else.

    Can I just use your data without having to download and host it myself?
    Probably, it depends on what you want to do. Shoot a PM and we'll talk! Our own website should be appearing as soon as we finalise our hosting.

    When will it be available?
    Soon; hopefully the next few days. We'll post a new thread in this forum as soon as we have the data in the right format - we need to break it down into weekly dumps and do some other minor formatting.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •