I am trying to port my little url shortener kurz.go (https://github.com/fs111/kurz.go), which uses godis to the latest weekly of golang and therefore I have to re-install all packages, that I use.

When I try to install godis, I get the following error:

$ go get github.com/simonz05/godis

github.com/simonz05/godis

../go/src/pkg/github.com/simonz05/godis/godis.go:78: too few values in struct initializer

What am I missing?

I'm probably overlooking something, but I don't see any way to close a client and release the connections it's holding.

This would be particularly important for pipeline clients where several might be opened simultaneously.

I'm trying to install godis on Go 1 Windows amd64, but I dunno why (maybe the tags) I'm not getting the correct repo from github.

package github.com/simonz05/godis/redis
        imports github.com/simonz05/godis/redis
        imports github.com/simonz05/godis/redis: import "github.com/simonz05/godis/redis": cannot find package

I see that the repo being pulled in my GOPATH does not have the same structure than the current master on github (maybe because the Go1 version pulls the Go1 tag).

Maybe fix tag?

Hi,

I'm trying to package godis for Arch Linux. I'm on a 64-bit computer, with go version "go1.0.2". After moving the examples directory away and compiling, the tests give me this:

=== RUN TestServer --- FAIL: TestServer (0.00 seconds) godis_test.go:15: monitor expected <nil> got <nil>, err(Connection error 127.0.0.1:6379) godis_test.go:15: requirepass foobared expected <nil> got <nil>, err(Connection error 127.0.0.1:6379) godis_test.go:15: ping authenticated expected <nil> got <nil>, err(Connection error 127.0.0.1:6379) panic: runtime error: invalid memory address or nil pointer dereference [recovered] panic: runtime error: invalid memory address or nil pointer dereference [signal 0xb code=0x1 addr=0x30 pc=0x42e14a]

goroutine 13 [running]: testing._func_003(0x7f88ab033fa8, 0x7f88ab033100, 0x7f88ab033fb8, 0x7f88ab033d18) /usr/lib/go/src/pkg/testing/testing.go:268 +0xf3 ----- stack segment boundary ----- github.com/simonz05/godis.(_Reply).StringArray(0x0, 0x0, 0x7f88ab033e70, 0x200000002) /tmp/yaourt-tmp-alexander/aur-go-godis/src/src/github.com/simonz05/godis/conn.go:134 +0x20 github.com/simonz05/godis.(_Reply).StringMap(0x0, 0x566b34, 0x757165720000000b, 0x0) /tmp/yaourt-tmp-alexander/aur-go-godis/src/src/github.com/simonz05/godis/conn.go:155 +0x25 github.com/simonz05/godis.TestServer(0xf840000540, 0xbc05ca0) /tmp/yaourt-tmp-alexander/aur-go-godis/src/src/github.com/simonz05/godis/commands_test.go:647 +0x2a6 testing.tRunner(0xf840000540, 0x636be0, 0x0, 0x0) /usr/lib/go/src/pkg/testing/testing.go:273 +0x6f created by testing.RunTests /usr/lib/go/src/pkg/testing/testing.go:349 +0x782

goroutine 1 [chan receive]: testing.RunTests(0x400c00, 0x636b08, 0x1600000016, 0x0, 0x627698, ...) /usr/lib/go/src/pkg/testing/testing.go:350 +0x7a5 testing.Main(0x400c00, 0x636b08, 0x1600000016, 0x6355f8, 0x300000003, ...) /usr/lib/go/src/pkg/testing/testing.go:285 +0x7a main.main() /tmp/go-build230035416/github.com/simonz05/godis/_test/_testmain.go:91 +0x91

goroutine 2 [syscall]: created by runtime.main /home/alexander/archpackages/community/go/trunk/src/go/src/pkg/runtime/proc.c:221

goroutine 4 [syscall]: syscall.Syscall6() /home/alexander/archpackages/community/go/trunk/src/go/src/pkg/syscall/asm_linux_amd64.s:40 +0x5 syscall.EpollWait(0xf800000006, 0xf84006d170, 0xa0000000a, 0xffffffff, 0xc, ...) /usr/lib/go/src/pkg/syscall/zerrors_linux_amd64.go:1781 +0xa1 net.(_pollster).WaitFD(0xf84006d160, 0xf84004c2c0, 0x0, 0x0, 0x0, ...) /usr/lib/go/src/pkg/net/fd_linux.go:146 +0x110 net.(_pollServer).Run(0xf84004c2c0, 0x0) /usr/lib/go/src/pkg/net/fd.go:236 +0xe4 created by net.newPollServer /usr/lib/go/src/pkg/net/newpollserver.go:35 +0x382 exit status 2 FAIL github.com/simonz05/godis 0.008s

Here's the package, if you're on 64-bit Arch Linux and want an easy way to reproduce the issue: https://aur.archlinux.org/packages.php?ID=61009

Get() unconditionally returns an error when called within a Multi() context. Modifying the code for Get() to check for the existence of a transaction still doesn't result in a GET command being emitted to Redis.

I suspect most of the data retrieval methods (those that don't return *Result) have similar issues, where they don't emit Redis commands.

When I try to connect to database != 0 with authentication. I must call Auth command before Select command.

Simple code that should work after my change:

---- begin of test.go ---- package main

import ( "godis" )

func main() { db := godis.New("", 1, "secret")

foo, err := db.Get("foo")
if err == nil {
    println(foo.String())
} else {
    println(err.String())
}

} ---- end of test.go ----

Hello,

First off, thanks for contributing the library. It appears to work pretty well.

If you call New with bad netaddr (i.e. 127.0.0.1 instead of 127.0.0.1:6379) then you will get a panic.

Here is the snippet:

na := strings.SplitN(netaddr, ":", 2) return &Client{Addr: na[1], Db: db, Password: password, net: na[0], pool: newPool()}

If there is not a colon in the addr passed then an index error occurs.

I'm not sure what the best mechanism is to address. You could check the length like this:

if len(na) != 2 { return nil }

Or you could check before calling split and do the same with strings.IndexRune(netaddr, ":").

Or you could make sure that it a valid net addr by doing a net.ResolveTCPAddr("tcp4", netaddr) and catch it on the error.

Was going to fork this and send a pull request just didn't wanted to do so without getting a stylistic blessing.

Thanks again!

Shelby

I'm polling a local Redis 2.6 server with:

client := redis.NewClient("localhost", 0, "")
reply, err := client.Call("INFO")
if err != nil {
    return ""
}
client.Call("QUIT")
return reply.Elem.String()

(And the equivalent with the non-exp package)

Occasionally, reply.Elem.String() will return truncated output:

"# Server\r\nredis_version:2.5.10\r\nredis_git_sha1:ac2febdb\r\nredis_git_dirty:0\r\nos:Linux 3.2.0-24-virtual x86_64\r\narch_bits:64\r\nmultiplexing_api:epoll\r\ngcc_version:4.6.3\r\nprocess_id:776\r\nrun_id:bdf2888b2974bccff9db6eb473abeb2ff2589d54\r\ntcp_port:10030\r\nuptime_in_seconds:14110158\r\nuptime_in_days:163\r\nlru_clock:1572223\r\n\r\n# Clients\r\nconnected_clients:5\r\nclient_longest_output_list:0\r\nclient_biggest_input_buf:0\r\nblocked_clients:0\r\n\r\n# Memory\r\nused_memory:941320\r\nused_memory_human:919.26K\r\nused_memory_rss:6828032\r\nused_memory_peak:104915264\r\nused_memory_peak_human:100.05M\r\nused_memory_lua:30720\r\nmem_fragmentation_ratio:7.25\r\nmem_allocator:jemalloc-3.0.0\r\n\r\n# Persistence\r\nloading:0\r\nrdb_changes_since_last_save:75038\r\nrdb_bgsave_in_progress:0\r\nrdb_last_save_time:1357899518\r\nrdb_last_bgsave_status:ok\r\nrdb_last_bgsave_time_sec:0\r\nrdb_current_bgsave_time_sec:-1\r\naof_enabled:0\r\naof_rewrite_in_progress:0\r\naof_rewrite_scheduled:0\r\naof_last_rewrite_time_sec:-1\r\naof_current_rewrite_time_sec:-1\r\n\r\n# Stats\r\ntotal_connections_received:440713\r\ntotal_commands_processed:15689002680\r\ninstantaneous_ops_per_sec:13878\r\nrejected_connections:0\r\nexpired_keys:0\r\nevicted_keys:0\r\nkeyspace_hits:2\r\nkeyspace_misses:95626\r\npubsub_channels:0\r\npubsub_patterns:0\r\nlatest_fork_usec:2982\r\n\r\n# Replication\r\nrole:master\r\nconnected_slaves:0\r\n\r\n# CPU\r\nused_cpu_sys:53159.38\r\nused_cpu_user:92844.89\r\nused_cpu_sys_children:828.06\r\nused_cpu_user_children:1923.70\r\n\r\n# Keyspace\r\n"

This happens using both exp and regular redis.

the install chapter

Importing godis to your code can be done with import "github.com/simonz05/godis". Thats it!

import "github.com/simonz05/godis" to import "github.com/simonz05/godis/redis"

Hi,

This pull request addresses three issues:

1. When a MULTI..EXEC transaction fails because of a WATCH condition, an error is returned by ReadAll. It is difficult to test this error to see if it was due to an aborted transaction or something else. I have named the error as ErrMultiAborted, so now you can just do a check like so:

   if err == redis.ErrMultiAborted { // transaction aborted, try again

2. The Call method of an AsyncClient cannot return any errors, so I removed the err return value. It calls Write in bytes.Buffer, but the docs for that method explicitly say it will never return an err (it just has the return type to match the io.Writer interface). Removing this lets careful people know it is okay to assume the call will succeed.

3. When using WATCH with MULTI..EXEC transactions, the normal pattern is:

   WATCH foo GET foo MULTI do stuff using the value of foo that is now guaranteed to not change EXEC

AsyncClient handles the MULTI..EXEC part great, but it is a bit verbose to run the commands that happen before the MULTI is issued. This adds a SyncCall to AsyncClient that works more like Call from Client. It fails if there are already commands in the queue.

Thanks,

Russ