Development

Well, this Saturday and Sunday is not for rest, because we have a 7-days’ vacation from October 1.

I think FreeBSD 6.0-STABLE (RELENG_6) is in a quite nice shape and it’s a good time to cut 6.0-RC1 (also RELENG_6_0) and be more careful committing code against RELENG_6. However, according to my diff service there are still a lot of unmerged changes.

I have submitted a plenty (about 4 or 5 or even 6 dozens) of MFC requests last week. I wish that the diff reduction part of these changes, in conjunction with important bugfixes, would got approved soon so we will have a healthy RC1 for test.

本周应该差不多了吧，不知道会有多少个RC哦……

不到4分钟就有人反对我提高某一默认值的提议，不过我还是在游说这样做的好处。

Will it provide better performance? Interested but not tested.

pav@ has put together this which looks quite interesting. Have a read of it :-)

(Got that from arved@’s blog)

I have an idea: let we have “a” is the number to determine.

return ( (0 == (a & (a-1))) && (a > 0))

Despite the style of the code, is there some better solution? Note that a is a native integer type (say, int, long, long long, etc).

Apparantly des@ has pointed out a serious problem with my patchset, say, the inconsistency of underlying API is the root of evil(tm), and should be corrected.

Let’s go go go!

So I got this in libmsun’s bsdsrc :-)

It’s not quite easy to represent a type that is found in base “template class”. Consider the following code:

template <class Key,class T>     class CMyClass: public map<Key,T>{public:    void lookup(Key x)    {        map<Key,T>::iterator iter;         iter = find(x);    }};

I started the work almost four months ago, but I have lost all my work last month. Therefore, I’m trying to integrate as much as possible whenever possible. The current workflow is:

• Bump WARNS
• Fix errors
• Testdrive on almost all T1 platform I can touch
• Commit