Date: Wed, 13 Jun 2007 08:26:38 -0700 From: Garrett Cooper <youshi10@u.washington.edu> To: Danny Braniss <danny@cs.huji.ac.il> Cc: hackers@freebsd.org, David Wolfskill <david@catwhisker.org> Subject: Re: Using shell commands versus C equivalents Message-ID: <46700CAE.6020902@u.washington.edu> In-Reply-To: <E1HySxb-000PIg-89@cs1.cs.huji.ac.il> References: <466F86C6.7010006@u.washington.edu> <20070613123213.GE98927@bunrab.catwhisker.org> <E1HySxb-000PIg-89@cs1.cs.huji.ac.il>
next in thread | previous in thread | raw e-mail | index | archive | help
Danny Braniss wrote:
>> --Rgf3q3z9SdmXC6oT
>> Content-Type: text/plain; charset=us-ascii
>> Content-Disposition: inline
>> Content-Transfer-Encoding: quoted-printable
>>
>> On Tue, Jun 12, 2007 at 10:55:18PM -0700, Garrett Cooper wrote:
>>
>>> Another simple question (I hope):
>>> Is there any reason why shell commands should be used in place of a=20
>>> C command (in this case chmod via vsystem instead of the chmod(2)=20
>>> function)? It seems like the fork / exec would be more expensive with=20
>>> the shell command, but any area where code could be optimized is more=20
>>> than welcome I would think.
>>>
>> There often are reasons to prefer using shell commands to C.
>>
>> There typically exist many tradeoffs involved in evaluating one over the
>> other, and "machine efficiency" is not always the highest goal. (For
>> example, it may be better to reduce complexity in favor of having a
>> simpler structure that is easier for people to understand and maintain
>> with confidence that the changes they make have the desired results.
>> This is, of course, not to try to claim that shell scripts are
>> inherently easier to understand than C code; that would be a silly
>> stance to take.)
>>
>> I commend to your attention Geoff Collyer and Henry Spencer's "C News"
>> (a successor to Rick Adams' "B News") implementation, a great deal of
>> which was written as shell scripts (ca. 1988 or so).
>>
>> (Yes, I realize that that was almost 20 years ago, and that it
>> didn't involve FreeBSD per se. Ignoring the lessons of history is
>> rather short-sighted and foolish: despite using shell scripts for
>> so much of the "code," the machine I was then running went from
>> being extremely busy all the time to having a couple of bursts of
>> activity per day for about an hour each time -- with more news
>> flowing with C News vs. B News.)
>>
>>
>
> read the question again, though it is not absolutely clear/correct, the question
> was:
> chmod(path, mode)
> vs
> system("chmod ...")
>
> and not wheather to write a program or a shell script.
>
> danny
>
>
>
Sorry -- actually I meant that (along similar lines), there was a
program with the following lines:
vsystem("/bin/chmod +x %s", filename);
and I replaced it with:
chmod(filename, (mode_t) ( S_IXUSR | S_IXGRP | S_IXOTH ));
Probably won't yield much gain overall, but every drop counts and there
are quite a few iterations performed in the pkg_* programs, in
particular dealing with X.org.
Next step, eliminating the linked list structure in favor or red-black
trees, maybe.
-Garrett
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?46700CAE.6020902>