Easynet. SOLR Development History

This morning, I suddenly wanted to write my own development process of easynet. SOLR. During this period, I encountered many problems, such as architecture problems and performance problems. I would like to share with you some of my ideas. I don't think there is any experience or something inappropriate. You are welcome to make a picture.

I remember that I decided to use SOLR for the company's full-text retrieval application. SOLR initially gave me the feeling that the function was perfect, and the performance also met the current needs. However, when I select the. NET client, I am a little embarrassed. At that time, there were only two options: solrsharp and SOLR. net. Solrsharp has not been updated for a long time, so it was decided to use SOLR. net. Because I like simplicity (I like simplicity, and I am a Fan Ke), I have simplified it in SOLR. net. I remember that on the evening of the official launch, I found some performance problems with SOLR. Net during online testing. Later, according to the dump file, it was mainly caused by serialization. So that night I changed serialization to interface-based implementation. After going online, we gradually optimized SOLR. NET and modified the XML-based protocol to the JSON-based protocol. Later, we sawCodeFinally, it was modified to adopt the javabin protocol officially recommended by SOLR.

To make it easier for other users to use SOLR on the Windows platform, I opened the SOLR. Net version easynet. SOLR that I changed myself last year. In subsequent days, the easynet. SOLR architecture was gradually modified to make it simpler and easier to expand. Currently, most SOLR functions are supported in released versions.

I have a principle: I provide the best method you and I think, but if you are not satisfied, you can expand on this architecture to implement your own ideas. Based on this principle, easynet. SOLR maintains the default javabin protocol and interface-based serialization implementation. However, by implementing icodec and icodecfactory, You can process XML, JSON, and other protocols. The general DNS interface isolrresponseparser provides full control over SOLR operation results.

It is difficult to build an open-source project. Sometimes, suddenly there is a new idea, and I may get up late at night to encode and implement it. One of my colleagues commented on easynet. SOLR on codeplex.

At the beginning, we should stick to it as much as possible, do as well as do better. I am very simple. I like open source, I like sharing, I like coding, and I am a Fan Ke.