java覆盖率测试工具？ - 未名空间MITBBS历史存档

国际科技财经博客移民网络热点娱乐民生时事公众号

Redian新闻

>未名空间

>Java - 爪哇娇娃

java覆盖率测试工具？

java覆盖率测试工具？# Java - 爪哇娇娃

v*s2005-08-17 07:08

1 楼

各位：不知道哪位有知道java下哪种覆盖率测试工具好用又是free的？
我用过emma, jcoverage, clover.
前两个是免费，但是效率比较低，插装class文件的。
第3个是commercial的，但是效率很高，插装src code。
还有其他工具推荐么？

g*g2005-08-17 07:08

2 楼

Rational code coverage is not bad, not free though.

【在 v*s 的大作中提到】

: 各位：不知道哪位有知道java下哪种覆盖率测试工具好用又是free的？
: 我用过emma, jcoverage, clover.
: 前两个是免费，但是效率比较低，插装class文件的。
: 第3个是commercial的，但是效率很高，插装src code。
: 还有其他工具推荐么？

m*t2005-08-17 07:08

3 楼

GroboUtils is another popular open source coverage tool. Although it
instruments byte code as well, so if you don't like emma, you probably
won't like GroboUtils either.
But frankly, tools that instrument byte code are generally better
than those work at source level because they are less intrusive and higher
compatibility. Besides, I don't see why byte code instrumenters are
inherently insufficient - from my experience with emma or GroboUtils, they
seem to be comparable to clover.
Even there

【在 v*s 的大作中提到】

m*t2005-08-17 07:08

4 楼

Of course my fat fingers were trying to type "inefficient".

【在 m******t 的大作中提到】

:
: GroboUtils is another popular open source coverage tool. Although it
: instruments byte code as well, so if you don't like emma, you probably
: won't like GroboUtils either.
: But frankly, tools that instrument byte code are generally better
: than those work at source level because they are less intrusive and higher
: compatibility. Besides, I don't see why byte code instrumenters are
: inherently insufficient - from my experience with emma or GroboUtils, they
: seem to be comparable to clover.
: Even there

g*g2005-08-17 07:08

5 楼

coverage is considered an offline operation and
efficiency is not very important.

【在 v*s 的大作中提到】

m*t2005-08-17 07:08

6 楼

Coverage is usually enabled during unit testing rather than system
testing. The part that's supposed to enforce the timeout value (i.e.
the client?) and the part that's supposed to get back by the timeout
value (i.e. the server?) should probably be tested separately in
unit testing.

【在 v*s 的大作中提到】