Re: [PATCH] git-daemon: Simplify child management and associated logging by

From: Stephen R. van den Berg <srb@cuci.nl>
Date: Aug 13, 2008, 07:36 UTC
Message-ID: <20080813073613.GB12628@cuci.nl>
In-Reply-To: <7vvdy5pxz6.fsf@gitster.siamese.dyndns.org>

Junio C Hamano wrote:

>"Stephen R. van den Berg" <srb@cuci.nl> writes:

>> Separating it causes two things:
>> a. The patches to become dependent on each other in the timeline.
>> b. More (redundant) work, because some parts that need to be rewritten, get
>>    deleted by the following patch(es).

>These are actually desirable properties from reviewability point of view.

Ok, I'll look into the splitup.

>>> - I see you have a call to vsyslog, which is the first user of the
>>>   function.  How portable is it (the patch coming from you, I know
>>>   Solaris would have it, and recent 4BSD also would, but what about the
>>>   others)?

>> Cygwin has it, Solaris does, Linux does, MacOSX does.
>> AIX and HPUX don't, perhaps.
>> I'll see what I can do to avoid it, yet simplify the code.

>That's one of the reasons why I asked you to split it to three patches, so
>that the syslog change can potentially be independently replaced with a
>better alternative.

Well, I've already found an alternative, and looks a lot more appealing than the buffer-juggling code of before.

>In any case, it is already late in the rc cycle; I'd like to apply your
>earlier "In SysV, signal(SIGCHLD) need to be rearmed" patch and nothing
>else for now.  The clean-up is very attractive but can be done post 1.6.0.

No problem. That's the reason I chipped that first patch off; it's a direct bugfix. And BTW, don't mistake me for a "Solaris guy". I've been using Linux almost exclusively since 1991. My SunOS/Solaris knowledge has been fading steadily ever since. It's just that I had to install git-daemon on someone else's Solaris box just recently.

-- 
Sincerely,
           Stephen R. van den Berg.

"And now for something *completely* different!"