Sumo DB and computing

[thatsumoguy 21]

On 05/11/2024 at 06:28, Octofuji said:

@thatsumoguy thanks again for this great resource :)

A bug seems to have appeared in the Rikishi Matches API, I noticed today when compiling Makuuchi consecutive tracker streaks (it was fine in September).

For quite a few rikishi there is a rogue Makushita result being inserted between on Day 1 of 2024-09.

For example for Tamawashi the 16th match is actually a Makushita match between two unrelated rikishi (ID 14 appears for both Tamawashi and Wakanosho):

    {
      "bashoId": "202409",
      "division": "Makuuchi",
      "day": 1,
      "matchNo": 8,
      "eastId": 14,
      "eastShikona": "Tamawashi",
      "eastRank": "Maegashira 10 East",
      "westId": 55,
      "westShikona": "Roga",
      "westRank": "Maegashira 10 West",
      "kimarite": "shitatedashinage",
      "winnerId": 55,
      "winnerEn": "Roga",
      "winnerJp": "狼雅"
    },
    {
      "bashoId": "202409",
      "division": "Makushita",
      "day": 1,
      "matchNo": 24,
      "eastId": 45,
      "eastShikona": "Toshunryu",
      "eastRank": "Yokozuna 1 East",
      "westId": 14,
      "westShikona": "Wakanosho",
      "westRank": "Maegashira 10 East",
      "kimarite": "oshidashi",
      "winnerId": 14,
      "winnerEn": "Wakanosho",
      "winnerJp": "若ノ勝"
    },

The same is true for quite a few others e.g. Daieisho, Meisei, Nishikigi, etc...

 

Oh thanks for pointing this out. I will check on it. There was a bug due to "reasons" in september, but I thought I'd caught all the issues.

[thatsumoguy 21]

@OctofujiI believe I've resolved this issue and all the erroneous matches have been removed.

 

[Octofuji 349]

On 25/11/2024 at 01:47, thatsumoguy said:

@OctofujiI believe I've resolved this issue and all the erroneous matches have been removed.

 

Thank you that's fixed for most of them I think.

Tamawashi's record still has Wakanosho in there (the 31st match now we've had another basho). I guess there are still some other "reasons"

https://www.sumo-api.com/api/rikishi/14/matches

(Also Takarafuji has Fukai interloping https://www.sumo-api.com/api/rikishi/25/matches)

The others in Makuuchi look fine.

Edited November 28 by Octofuji
Add Takarafuji/Fukai

[Octofuji 349]

Out of interest @thatsumoguy, is the source code for your API available anywhere (I couldn't find it on GitHub).

I don't know anything about Go, but if I was ever going to learn that should definitely be the place I'd start!

[thatsumoguy 21]

On 29/11/2024 at 06:13, Octofuji said:

Thank you that's fixed for most of them I think.

Tamawashi's record still has Wakanosho in there (the 31st match now we've had another basho). I guess there are still some other "reasons"

https://www.sumo-api.com/api/rikishi/14/matches

(Also Takarafuji has Fukai interloping https://www.sumo-api.com/api/rikishi/25/matches)

The others in Makuuchi look fine.

Thanks, those are gone now. Basically I moved hosting, forgetting that I use a cache for fast lookup of rikishis by the ID that the JSA provide. I forgot to transfer the cache over, and so it started reassigning the ids. I realised during day 1 and fixed it, and I thought i'd removed all the erroneous matches, but apparently not.

[Yamanashi 3,776]

2 hours ago, thatsumoguy said:

Thanks, those are gone now. Basically I moved hosting, forgetting that I use a cache for fast lookup of rikishis by the ID that the JSA provide. I forgot to transfer the cache over, and so it started reassigning the ids. I realised during day 1 and fixed it, and I thought i'd removed all the erroneous matches, but apparently not.

It's always the brilliant time-saving ideas that sometimes end up wasting the most time, isn't it?

[thatsumoguy 21]

21 hours ago, Yamanashi said:

It's always the brilliant time-saving ideas that sometimes end up wasting the most time, isn't it?

Often. The project was mostly about combining my enjoyment of sumo and trying to use different technologies. The cache is definitely not required, the db would be fast enough by itself, given the load.