Replace StringBuffer with StringBuilder #243
Conversation
…der where applicable - replacement is done when StringBuffer is used as a local variable - public APIs using StringBuffer are kept but deprecated in favor of the same using StringBuilder (note a method name change is required not to prevent code using null as parameter to compile)
|
First off I think this is awesome. I've been thinking about this every time I poke through the code. Nice to see someone got around to it. I don't see a point to maintaining two copies of the public methods. They may be public, but they're still internal to the driver. The only true public API of the driver is the core JDBC API and the extensions available via casting to PGConnection (ex: the CopyManager API). Nobody should be using the "internal" public methods (ex: the Utils class) and, if they are, it'd be their responsibility to update their code to use a StringBuilder instead. I say this as someone who has used those internal helper functions in the past. This is especially true if it'll be done as a new major release of the driver. Users that are upgrading to a 9.4 (or I guess 9.5?) driver can be expected to compile/test it with their app. It'll immediately be apparent what they need to fix. @davecramer @ringerc : What do you guys think? |
|
Ya, I see no reason to keep the old methods around. Dave Cramer On 5 January 2015 at 21:24, Sehrope Sarkuni notifications@github.com
|
|
Just note that the real logic is done in a shared private method. Public methods just delegate the job to it whether the provided arg is a StringBuilder or StringBuffer. So there not really two copies to maintain actually. |
Replace StringBuffer with StringBuilder for performance
Take advantage of Java 5 using StringBuilder where applicable
StringBufferis used as a local variableStringBufferare kept but deprecated in favor of the same usingStringBuilder. Cases can be found inorg.postgresql.core.Utils. New public methods usingStringBuilderdo not use the same name as theirStringBuffercounterpart otherwise it would prevent existing code usingnullas a method argument to compile.