Shaya Potter

the reason for this (per a cursory analysis) is that the directory is written (at add time) to the queue file (queue/queue). i wonder if a sed could fix it?...

thoughts about changing this so its not stored at enqueue time but determined at dequeue time?

should note, that I have a weirdness in my test module code for handling the error case, if I don't do that and just do a RM_ReplyWithCallReply for everything (i.e....

the above test / module implementation is problematic, as the docs for setting acl in redis module code is not quite accurate (as I learned when I tried to do...

Open Q: currently RM_CallWithUser will accept a NULL RedisModuleUser (and hence be equivalent to a plain RM_Call(). Should this be an error or is that behavior the desired behavior. i.e....

@oranagra can you review this?

So in summary of above, I reverted all the refactor changed to RM_Call. No More RM_CallWithUser. now we have a RM_SetContextModuleUser. If set (open Q above if it should be...

sorry for the rebase, I was trying to squash to additional commits together and that rebased my whole branch by accident.

yeah, I meant to just squash some local commits together before pushing, but muscle memory was rebase -i upstream/unstable (need to break that muscle memory). And then after realizing push...

In private conversation @MeirShpilraien wondered why we need the new `RM_SetContextModuleUser` API when we have `RM_AuthenticateClientWithUser`. It works by overwriting the user of `ctx->client->user` with whatever the caller passed in....