Similar to #245, MBULK reply containing an empty string can not be properly decoded.
Reply.scala and related tests
fixed bug(#244): "" -> "$0\r\n\r\n" instead of "$-1\r\n" according to h…
And, the error reply of incorrect data type switched from 'ERR' to 'WRONGTYPE' in test
removed an extra blank line
Removed the redundant case 0 branch
Removed the case 0 pattern matching branch for empty string which is redundant in function decodeBulkReply
Fixed a test to temporarily support redis server 2.8
In Redis 2.8, Many errors are now prefixed by a more specific error code instead of the generic -ERR, for example -WRONGTYPE, -NOAUTH, ...
this will be a little faster if we use endsWith instead of contains
removed extra spaces
use endsWith instead of contains
Fixed the empty string issue of MBULK_REPLY
Although it's OK to do these two pull requests concurrently, it will be confusing if we try to apply the same commits again. We shouldn't have to repeat commits. Could you make a new pull request which doesn't include the repeat commits? One way of doing this would be to make a patch that includes only the changes from ad2d291 and applying that to the current master branch and making that the branch for 246.
Could you also elaborate on the results of this change and what precisely the modifications entail?
ad2d291 does depend on changes in ca3b856 to pass all the tests:
1. BulkReply needs to be changed to allow having empty message be encoded and decoded
2. The WRONGTYPE error message thing
WRONGTYPE error message
I will say that let's just wait for #245 to be finally merged in master.
Got it. Good idea.
Out-of-date. see #251