Re: [Chicken-users] Redis-client issues

[Carl Douglas]

Hi Matt,

Thanks for reporting these bugs! 

 I am overseas on holiday at the moment and the soonest I will be able to take 
a look is next week after the 26. 

Regardless, you are welcome to fix these yourself, and I would encourage it! 

The redis-client.scm code is minimal and does not specialise per redis command.

Also you may like to consider an alternative redis implementation for chicken 
scheme, there is one hosted on gitorious.org somewhere I believe. 

Nevertheless I would like to see improvements made to the redis-client egg and 
intend to take a look at the problems you mention when I get back.

Thanks,
Carl



On 20/08/2012, at 2:23, Matt Gushee <address@hidden> wrote:

> Hi, Chickeneers--
> 
> I'm back again! Though I have never been a key member of this community, so 
> you are forgiven for not remembering me ;-)
> 
> Anyway, I am planning to develop a rather ambitious project using Redis, and 
> have been experimenting with the redis-client egg. Thanks to Carl Douglas for 
> developing it! But I have noticed a couple of minor problems with the egg; 
> one at least is surely a bug. Oh, and I am using version 0.7 of the egg with 
> Chicken 4.7.0 & Redis 2.4.15 on Arch Linux.
> 
> Issue #1
> -----------
> 
> Nil responses are not handled in some cases, e.g.:
> 
>   csi> (redis-hset "goodies" "wine" "merlot")
>   (1)
>   csi> (redis-hset "goodies" "cheese" "gouda")
>   (1)
>   csi> (redis-hget "goodies" "wine")
>   ("merlot")
>   csi> (redis-hget "goodies" "cheese")
>   ("gouda")
>   csi> (redis-hget "goodies" "fruit")       ; Querying a non-existent field
>   ; never returns
> 
> When I do this in the command-line client (redis-cli), the response is (nil) 
> . At any rate, it is handled gracefully. Should be in our library, too, I 
> would think.
> 
> 
> Issue #2
> -----------
> 
> I don't know if this is really a bug, or just a feature I don't like, but if 
> a Scheme error (such as invalid input) is raised in the midst of a 
> transaction, the client seems to be rendered useless.
> 
>   csi> (redis-multi)
>   ("OK")
>   csi> (redis-sadd "abc" "def")
>   ("QUEUED")
>   csi> (redis-hset "xyz" '())
> 
>   Error: (string-length) bad argument type: ()
> 
>         Call history:
> 
>         <syntax>          (redis-hset "xyz" (quote ()))
>         <syntax>          (quote ())
>         <syntax>          (##core#quote ())
>         <eval>    (redis-hset "xyz" (quote ())) <--
>   csi> (redis-sismember "abc" "def")
>   ("QUEUED")
>   csi> (redis-lpush "fubar" "baz")
>   ("QUEUED")
>   csi> (redis-lpop "fubar")
>   ("QUEUED")
>   
> In other words, any further commands simply produce the response ("QUEUED"). 
> Now, I have found that I can write an exception handler that calls 
> (redis-discard), and that seems to solve the problem. But what would make the 
> most sense to me is to build an exception handler into the client library 
> that would discard the transaction and then re-raise the exception. What says 
> the community?
> 
> Also ... though I'm far from an expert Schemer, if Carl is not available to 
> work on this stuff, I will attempt to fix the egg, and if I am able to do so, 
> I'd be happy to contribute my fixes to the community.
> 
> --
> Matt Gushee
> _______________________________________________
> Chicken-users mailing list
> address@hidden
> https://lists.nongnu.org/mailman/listinfo/chicken-users