ActiveRecord findall pool duplicating keys.

[Toorion]

RPUSH is saved keys to findall pool with duplicated values. Because Redis LIST is not unique. Finally find()->all() return a lot of duplicated objects. As solution it can use sorted set and store objects with SADD and get it with SMEMBERS command instead of LRANGE.

[vaseninm]

@cebe I can fix it. But I'm not sure that's the right way. I think that when you duplicate PK to throw an exception.

[cebe]

agree that it should not be possible to insert a duplicate key but it should not take too long to insert into big datasets. the check should be done fast somehow.

[vaseninm]

@cebe

EVAL "for i = 1, 1000000 do redis.call('SADD', 'test', i) end" 0            (7.90s)
SMEMBERS test               (6.75s)
EVAL "for i = 1, 1000000 do redis.call('RPUSH', 'test', i) end" 0         (7.66s)
LRANGE test 0 -1             (6.46s)

Not slow?

[cebe]

what do you mean with "not slow"? We can not change the storage format in 2.0.x btw...

[vaseninm]

Exactly. Lists do not have fast methods to check for existence.

[nailfor]

Hello, vaseninm see fork https://github.com/nailfor/yii2-redis After some optimizations, difference in speed between the YII / Redis and my fork is small

Storage pool is change the format of my realization