Date: Sat, 17 Nov 2007 17:39:10 -0500 From: Skip Ford <skip@menantico.com> To: "Bjoern A. Zeeb" <bzeeb-lists@lists.zabbadoz.net> Cc: Yuri <yuri@rawbw.com>, Robert Watson <rwatson@FreeBSD.org>, freebsd-hackers@FreeBSD.org Subject: Re: How to get filename of an open file descriptor Message-ID: <20071117223910.GD813@menantico.com> In-Reply-To: <20071117215003.U53707@maildrop.int.zabbadoz.net> References: <1194905125.4738ce25a968c@webmail.rawbw.com> <20071112222557.N81124@fledge.watson.org> <1194980181.4739f355a32bc@webmail.rawbw.com> <20071114104157.D92502@fledge.watson.org> <20071114112304.GA835@menantico.com> <20071114121812.U2025@fledge.watson.org> <20071114132743.GB835@menantico.com> <20071116144356.S10677@fledge.watson.org> <20071116212342.GD835@menantico.com> <20071117215003.U53707@maildrop.int.zabbadoz.net>
next in thread | previous in thread | raw e-mail | index | archive | help
Bjoern A. Zeeb wrote: > On Fri, 16 Nov 2007, Skip Ford wrote: > > >How about renaming procstat(1) to proc(1), rolling up all of > > calling it proc(1), I think, is actually not a good idea either. That > is way more confusing for people who still think about /proc and do > not know the difference between (1) or (4). > > I like the procstat as it aligns well with other things like > fstat netstat sockstat systat vmstat gstat iostat pmcstat ... > > I admit we also have some *info tools like ffsinfo/diskinfo/rpcinfo/.. > but ``pinfo'' seems to better fit the *stat category of tools;-) > > I am not able to find anything but a simple "C wrapper" for > /proc/*/stat for linux on the web easily (which I suppose could as well > be a sh skript) and cannot even find something like procstat on the > linux machines I have access to. But there seems to be a procinfo that > seems to as well extract information from /proc/ on linux. So having > pinfo or procinfo might more confuse people to expect something > differently and even worse might mean to be the same tool with > compatible command line. > > While thinking we should try to aling with other OSes and not confuse > users coming from non-BSD worlds, procstat to mee seems to be the > thing that would best fit for our tree. I don't mind the name procstat(1), I just think we already have one that happens to be abbreviated ps(1) instead of being spelled out. If we end up with hardlinks for a proc tool family of utilities, users will be pointed to the actual tool they need rather than proc(1) so I don't think new user confusion would be that great. But, the same argument also really nullifies my argument for the name as well. If we have hardlinks, I care much less about the name of the base utility since it won't be used everyday. With hardlinks, pinfo(1), proc(1), and procstat(1) are all fine with me. The OP in this thread would then just use pfiles(1) to get a list of open files, same as Solaris, no matter what we call it. It would be interesting to know for sure, though, if Solaris uses hardlinks and, if so, what their utility is called. -- Skip
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20071117223910.GD813>