Re: [recommenders-dev] Snipmatch GSOC

[Marcel Bruch <bruch@xxxxxxxxxxxxxxxxxx>]

Yes, let's see what Doug thinks about it.

Regarding the file names:

I think that file names don't carry any meaningful information. They just serve as "primary key" for the index. All information should be contained in the file.

Regarding directories:

One type of snippets

One directory per snippet type sounds ok. What "kind of snippets" do you have in mind?

Regarding XML format:

I'd stick to JSON since we use JSON all over the project. For de-serialization we use Google's GSON library. See GsonUtil class to get a first impression on how to use GSON.

On 20.02.2012, at 09:43, chen cheng wrote:

I agree with file system storage solution. Do not know what Doug's opinion is. 

But before coding, we must define some details:

1. One snippet one file, with XML format. Is it OK? Does file name make sense, or just keep them unique.

2. One type of snippets, one directory. Common repositories has its own directory,  Personal repositories has its own direcotry. Is it OK?

Thanks,

Marcel

-- 
Eclipse Code Recommenders:
 w www.eclipse.org/recommenders
 tw www.twitter.com/marcelbruch
 g+ www.gplus.to/marcelbruch