Sphinx 1.10-beta 测试版 新功能介绍

发布: 2010-08-28 22:09



2010年7月19日,Sphinx 1.10-beta版本发布,增加了很多新功能,主要亮点如下:



  • 实时索引支持
  • 字符串字段值存储
  • MPM模式支持
  • 外部文本文件索引
  • searchd直接支持MySQL协议
  • SphinxQL接口查询完善




新增功能列表:



  • 实时索引:added RT indexes support (Chapter 4, Real-time indexes)

  • MPM模式支持:fork、prefork、threads;added prefork and threads support (workers directives)

  • 本地分布式搜索支持多线程;added multi-threaded local searches in distributed indexes (dist_threads directive)

  • AddQuery()子查询缓存added common subquery cache (subtree_docs_cache,
    subtree_hits_cache directives)

  • 文本字段信息存储及返回;added string attributes support (sql_attr_string,
    sql_field_string,
    xml_attr_string,
    xml_field_string directives)

  • 索引时词汇数统计;added indexing-time word counter (sql_attr_str2wordcount,
    sql_field_str2wordcount directives)

  • SphinxQL模式支持高亮调用和分词结果调用;added CALL SNIPPETS(),
    CALL KEYWORDS() SphinxQL statements

  • SphinxQL模式支持字段权重和索引权重设置;added field_weights, index_weights options to
    SphinxQL SELECT statement

  • SphinxQL模式支持SphinxSE(仅insert数据);added insert-only SphinxQL-talking tables to SphinxSE (connection='sphinxql://host[:port]/index')

  • SphinxSE查询中支持select计算;added select option to SphinxSE queries

  • 增加searchd崩溃coredump文件设置;added backtrace on crash to searchd

  • 支持从SQL字段属性值调入外部文本文件索引;added SQL+FS indexing, aka loading files by names fetched from SQL
    (sql_file_field directive)

  • 多线程模式时支持watchdog守护;added a watchdog in threads mode to searchd

  • 索引合并时自动消除脏数据;added automatic row phantoms elimination to index merge

  • 不记录词汇在文档的位置(可减小索引大小和提高性能);added hitless indexing support (hitless_words directive)

  • indextool增加参数:added --check, --strip-path, --htmlstrip, --dumphitlist ... --wordid switches to indextool

  • searchd增加参数--stopwait提升退出时保存信息的能力,--logdebug设定调试信息输出;added --stopwait, --logdebug switches to searchd

  • indexer增加参数--dump-rows用以输出所有读取到的数据,--verbose用以输出详细的问题信息;added --dump-rows, --verbose switches to indexer

  • 混合字符设置(AT&T可被所分解为AT&T、AT、T);added "blended" characters indexing support (blend_chars directive)

  • 增加sql_joined_field,用于支持类似SQL的GROUP_CONCAT处理,以及对SPH_RANK_PROXIMITY_BM25下评分值的控制;added joined/payload field indexing (sql_joined_field directive)

  • 增加FlushAttributes()接口用于将内存信息保存到磁盘;added FlushAttributes() API call

  • BuildExcerpts增加多项处理参数;added query_mode, force_all_words, limit_passages, limit_words, start_passage_id, load_files, html_strip_mode,
    allow_empty options, and %PASSAGE_ID% macro in before_match, after_match options
    to BuildExcerpts() API call

  • SELECT支持获取@groupby/@count/@distinct信息;added @groupby/@count/@distinct columns support to SELECT (but not to expressions)

  • 查询时自动展开词汇(可能会降低性能);added query-time keyword expansion support (expand_keywords directive,
    SPH_RANK_SPH04 ranker)

  • 批量查询数目限制;added query batch size limit option (max_batch_queries directive; was hardcoded)

  • 增加SINT()函数用于正确解析32位整数到64位;added SINT() function to expressions



新功能说明:



  • 因为目前新增功能较多,相关功能还在监测和完善中,部分功能可能存在不可预知的问题,暂时请不要使用到正式场合!
  • 实时索引:目前限制较多,尚不太稳定,已知问题请查看问题列表
  • 字符串值存储:单字符串4MB大小限制,不要存储大量的数据,尚未优化更好的存储方式
  • 详细变更列表:查看


测试版文档:




测试版下载:























































    Sphinx 1.10-beta 新! (r2420; 2010-07-19)
    Source tarball (tar.gz) 1.10-beta 1.4M 下载
    Win32 binaries w/MySQL support 1.10-beta 3.1M 下载
    Win32 binaries w/MySQL+PostgreSQL support 1.10-beta 4.6M 下载
    Win32 binaries w/MySQL+PgSQL+libstemmer+id64 support 1.10-beta 4.9M 下载
    RHEL 5.x x86_64 RPM 1.10-beta 3.1M 下载
    RHEL 5.x i386 RPM 1.10-beta 3.6M 下载
    RHEL 4.x i386 RPM 1.10-beta 3.5M 下载
    RHEL 5.x x86_64 binaries 1.10-beta 3.1M 下载
    RHEL 5.x i386 binaries 1.10-beta 3.6M 下载
    RHEL 4.x i386 binaries 1.10-beta 3.5M 下载




原文: http://qtchina.tk/?q=node/503

Powered by zexport