Suggestion: rename "killall" to "fkill", but wait five years to phase the new name in
Jason A. Spiro
Dear hackers, thanks for maintaining it on FreeBSD.
Naming it the same as System V killall, which just kills all
processes, can wreak havoc. When someone types a standard Linux
killall command line as root on a Solaris or HP-UX server, System V
killall runs and kills all processes.
It might be good if you'd rename it to something else. Not "akill"
(All Kill): it looks like IRIX probably ships with something called
akill already, so this would be confusing. Maybe "fkill" (Friendly
Kill).
You could do this in phases: for the first five years,
/usr/bin/killall could print a warning onscreen, then function as
usual. After five years, it could cease to function unless you call
it as "fkill".
Craig, and hackers, are you both willing to do this?
-Jason
--
Jason Spiro: software/web developer, packager, trainer, IT consultant.
I support Linux, UNIX, Windows, and more. Contact me to discuss your needs.
+1 (416) 992-3445 / www.jspiro.com
_______________________________________________
freebsd...@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-hackers
To unsubscribe, send any mail to "freebsd-hacke...@freebsd.org"
Daniel O'Connor
> Naming it the same as System V killall, which just kills all
> processes, can wreak havoc. �When someone types a standard Linux
> killall command line as root on a Solaris or HP-UX server, System V
> killall runs and kills all processes.
>
> It might be good if you'd rename it to something else. �Not "akill"
> (All Kill): �it looks like IRIX probably ships with something called
> akill already, so this would be confusing. �Maybe "fkill" (Friendly
> Kill).
<snark>
Why not get Sun and HP to change killall to match Linux & *BSD
behaviour?
</snark>
Although seriously, why not? killall just killing everything is a fairly
dangerous command with almost no use in the real world.
--
Daniel O'Connor software and network engineer
for Genesis Software - http://www.gsoft.com.au
"The nice thing about standards is that there
are so many of them to choose from."
-- Andrew Tanenbaum
GPG Fingerprint - 5596 B766 97C0 0E94 4347 295E E593 DC20 7B3F CE8C
Xin LI
> Craig, and hackers, are you both willing to do this?
No.
killall is not part of standard, and, just because System V choose to
implement that way, does not warrant that FreeBSD has to. Moreover,
user can always alias /sbin/killall to 'fkill' and 'kill -15 -1' to
'killall' if they really want the System V behavior.
Cheers,
--
Xin LI <del...@delphij.net> http://www.delphij.net
Dag-Erling Smørgrav
DES
--
Dag-Erling Smørgrav - d...@des.no
Gary Jennejohn
Xin LI <del...@gmail.com> wrote:
> On Mon, Dec 21, 2009 at 10:31 PM, Jason A. Spiro <jason...@gmail.com> wrote:
> > Craig, and hackers, are you both willing to do this?
>
> No.
>
> killall is not part of standard, and, just because System V choose to
> implement that way, does not warrant that FreeBSD has to. Moreover,
> user can always alias /sbin/killall to 'fkill' and 'kill -15 -1' to
> 'killall' if they really want the System V behavior.
>
I'm wondering why we even need killall when pkill seems to have the same
basic functionality and is located in /bin (and /rescue) rather than /usr/bin?
---
Gary Jennejohn
Matthias Andree
> man pkill
And that one is also provided on Solaris.
--
Matthias Andree
Craig Small
> Naming it the same as System V killall, which just kills all
> processes, can wreak havoc. When someone types a standard Linux
> killall command line as root on a Solaris or HP-UX server, System V
> killall runs and kills all processes.
The problem for me is that killall in Linux has been called that for a
very long time now. psmisc came out 11 years ago and before that killall
was in procps. I'm not sure when but the copyright message says 1994.
That's 16 years or more of people getting very used to killall doing
what it does and being called killall. I know of the problem you refer
to having administered Solaris servers before, but changing the name now
will cause more problems than it solves.
> Craig, and hackers, are you both willing to do this?
I'm not. Even though I just got a new SATA drive its not big enough to
handle the torrent of emails from people saying "why did i do that" and
"who cares about Solaris" etc etc if I did change it.
Sounds like its a no from FreeBSD folk too. In fact to me its more
important my various programs look the same(ish) across the Linux
distributions and to FreeBSD than they are to Solaris.
I also agree with Daniel; why would anyone want to literally kill every
process?
- Craig
--
Craig Small GnuPG:1C1B D893 1418 2AF4 45EE 95CB C76C E5AC 12CA DFA5
http://www.enc.com.au/ csmall at : enc.com.au
http://www.debian.org/ Debian GNU/Linux, software should be Free
Nate Eldredge
> I also agree with Daniel; why would anyone want to literally kill every
> process?
AFAIK, it's a helper program for shutdown(8) (or shutdown(1M) as they call
it) and isn't really intended to be useful otherwise.
--
Nate Eldredge
na...@thatsmathematics.com
Stephen Montgomery-Smith
"screw-the-whole-system." It would do something like this:
while true; do \
echo "Please wait while your system is being destroyed..."
sleep 10
done
Jason Spiro
[snip]
> PS: Applying your changes to a mailing list are not const.
What does that sentence mean?
Jason Spiro
> I'm wondering why we even need killall when pkill seems to have the same
> basic functionality and is located in /bin (and /rescue) rather than /usr/bin?
I like killall because of its -v (verbose) option. It lets me know what killall
killed. You just inspired me to request the pkill upstream maintainer to add a
similar option. My feature request is at http://bugs.debian.org/562111 .
-Jason
Doug Barton
> jhell <jhell <at> DataIX.net> writes:
>
> [snip]
>> PS: Applying your changes to a mailing list are not const.
>
> What does that sentence mean?
My version of that would be:
1. More polite: You can easily alias killall='echo NO!' on your
solaris machines to avoid accidentally typing killall there.
2. Less polite: This is not FreeBSD's problem, go away. :)
Doug
--
Improve the effectiveness of your Internet presence with
a domain name makeover! http://SupersetSolutions.com/
Jason Spiro
> On Tue, Dec 22, 2009 at 01:31:02AM -0500, Jason A. Spiro wrote:
>
> Hello Jason (and the FreeBSD folk),
>
> The problem for me is that killall in Linux has been called that for a
> very long time now. psmisc came out 11 years ago and before that killall
> was in procps. I'm not sure when but the copyright message says 1994.
>
> That's 16 years or more of people getting very used to killall doing
> what it does and being called killall. I know of the problem you refer
> to having administered Solaris servers before, but changing the name now
> will cause more problems than it solves.
What problems will it cause, other than a torrent of complaints?
> > Craig, and hackers, are you both willing to do this?
>
> I'm not. Even though I just got a new SATA drive its not big enough to
> handle the torrent of emails from people saying "why did i do that" and
> "who cares about Solaris" etc etc if I did change it.
I can create a special email address for this, you can mention "Complaints to"
and the new address in the warning message, and I can try to reply to everyone.
I will make use of form letters whenever it makes sense to.
> Sounds like its a no from FreeBSD folk too. In fact to me its more
> important my various programs look the same(ish) across the Linux
> distributions and to FreeBSD than they are to Solaris.
[snip]
I will reply to Xin's "no" message later. I agree that it's a good idea that
you and the FreeBSD folk should use the same name for the same utility.
Jason Spiro
> <snark>
> Why not get Sun and HP to change killall to match Linux & *BSD
> behaviour?
> </snark>
>
> Although seriously, why not? killall just killing everything is a fairly
> dangerous command with almost no use in the real world.
Because I find that when I send feedback to closed-source software vendors, I
often get no reply at all.
But if any of you want to file an OpenSolaris bug or write to HP or any other of
the Unix vendors who ship a dangerous killall command (I think most do), feel
free. Please let us know that you've done so, to help us avoid duplication.
Jason A. Spiro
>
> No.
>
> killall is not part of standard, and, just because System V choose to
> implement that way, does not warrant that FreeBSD has to. �Moreover,
> user can always alias /sbin/killall to 'fkill' and 'kill -15 -1' to
> 'killall' if they really want the System V behavior.
Xin, I'd like to discuss this issue with you by some means other than
email. Could you please phone me at (646) 461-3412 (my Google Voice
call forwarding number, New York, NY); or is Skype voice chat okay?
(Those are my preferred options.)
Or how about IRC or MSN text discussion?
I can summarize the results of our conversation to the mailing list afterwards.
Xin LI
On Tue, Dec 22, 2009 at 2:06 PM, Jason Spiro <jason...@gmail.com> wrote:
> jhell <jhell <at> DataIX.net> writes:
>>
>> This is what shell aliases are for and what a system admins job consist
>> of. If it gives you that much of a problem just alias it out for your self
>> in your .cshrc .shrc .bashrc .bash_profile etc. If you want to change
>> something on a more per user basis figure out how to setup a skeleton
>> directory so when a new user is created they get all the files from that
>> skel copied into there home. If it is more of a system-wide change then
>> the shell files in /etc will probably be of more use.
>>
>> PS: Applying your changes to a mailing list are not const.
>
> Using aliases would help me, but wouldn't help people elsewhere in the world who
> don't know what SysV killall does. Renaming FreeBSD killall would help prevent
> them from getting burned, perhaps on a busy production server, even once.
I'm afraid that it's too late to change either parties, i.e. there
would be a lot of scripts that rely on the BSD or Linux behavior, etc.
Instead of making changes to killall which already diverge between
open source implementation and closed source ones, it might be better
off to have administrators to learn some more consistent ways to do
the same task, i.e. pkill.
Cheers,
--
Xin LI <del...@delphij.net> http://www.delphij.net
Jason A. Spiro
On Tue, Dec 22, 2009 at 5:34 PM, Xin LI <del...@gmail.com> wrote:
> I'm afraid that it's too late to change either parties, i.e. there
> would be a lot of scripts that rely on the BSD or Linux behavior, etc.
That is why I suggested that you first show a warning message for five
years, then do the renaming.
> �Instead of making changes to killall which already diverge between
> open source implementation and closed source ones,
If you rename the open source killall to "fkill", then you will no
longer have a killall command which differs between open source and
closed source.
> it might be better
> off to have administrators to learn some more consistent ways to do
> the same task, i.e. pkill.
It would be good if sysadmins learned not to use killall. But I think
that most sysadmins who are already used to killall are unlikely to
learn not to type the command "killall" unless you rename open-source
killall to a different name like "fkill".
I think it's impractical to expect all sysadmins to switch to pkill.
Pkill is missing the option which displays a list onscreen of which
processes were killed. I sent a feature request to the maintainer,
but there is no guarantee that the maintainer will add that option.
And maybe there are other pkill options which are missing from skill.
Cheers,
-Jason
Jason A. Spiro
[snip]
> Xin, I'd like to discuss this issue with you by some means other than
> email.
Followup to my earlier message: Thanks for sending me a private mail
with your Jabber address. I added you. But then I saw your most
recent list post, and realized that I'd prefer to discuss killall by
list posting than private instant messaging for now.
-Jason
Doug Barton
> Using aliases would help me, but wouldn't help people elsewhere in the world who
> don't know what SysV killall does.
Seriously, it's not our problem if solaris did something stupid. There
is no hope whatsoever that you're going to get every Unix that has a
rational 'killall' command to change, so can we please drop this thread?
Thanks,
Doug
--
Improve the effectiveness of your Internet presence with
a domain name makeover! http://SupersetSolutions.com/
_______________________________________________
Xin LI
> Hi Xin,
>
> On Tue, Dec 22, 2009 at 5:34 PM, Xin LI<del...@gmail.com> wrote:
>
>> I'm afraid that it's too late to change either parties, i.e. there
>> would be a lot of scripts that rely on the BSD or Linux behavior, etc.
>
> That is why I suggested that you first show a warning message for five
> years, then do the renaming.
killall can be used by scripts which just works in the past, and will
never notice the warnings. Also, killall is not "that" dangerous on
FreeBSD, we should ONLY give warnings when it's really necessary,
otherwise users would just ignore all warnings we gave to them.
On the other hand, it seems to us that warning messages won't work, no
matter how long we give it, it is being ignored by a majority of users.
>> Instead of making changes to killall which already diverge between
>> open source implementation and closed source ones,
>
> If you rename the open source killall to "fkill", then you will no
> longer have a killall command which differs between open source and
> closed source.
Then users are already familiar with FreeBSD would have to learn what
"fkill" is, and after all, having them to pay for mistakes made by
commercial Unix vendors does not seem to be a fair option.
>> it might be better
>> off to have administrators to learn some more consistent ways to do
>> the same task, i.e. pkill.
>
> It would be good if sysadmins learned not to use killall. But I think
> that most sysadmins who are already used to killall are unlikely to
> learn not to type the command "killall" unless you rename open-source
> killall to a different name like "fkill".
Well, I'd say it's too late for us to change since it's several years
after we have 'killall' our way.
> I think it's impractical to expect all sysadmins to switch to pkill.
> Pkill is missing the option which displays a list onscreen of which
> processes were killed. I sent a feature request to the maintainer,
> but there is no guarantee that the maintainer will add that option.
> And maybe there are other pkill options which are missing from skill.
pkill have '-I', at least on FreeBSD...
Cheers,
--
Xin LI <del...@delphij.net> http://www.delphij.net/
FreeBSD - The Power to Serve! Live free or die
Jille Timmermans
change it's name to "killall"
Stephen Montgomery-Smith schreef:
Jason A. Spiro
> Seriously, it's not our problem if solaris did something stupid.
Actually, it looks like the mistake was made by Linux and FreeBSD
developers. SunOS had[1] killall in 1992, and maybe earlier. Craig
said the earliest copyright date on Linux killall is 1994. I think
FreeBSD killall is newer than 1994, since its manpage says that it was
modeled after the killall included with other OSes. I think this must
mean Linux.
When Linux and FreeBSD made their kill-selected-processes command,
they shouldn't've called it killALL.
But mistakes happen. Luckily, mistakes can be corrected.
> There
> is no hope whatsoever that you're going to get every Unix that has a
> rational 'killall' command to change, so can we please drop this thread?
It's never too late to change something for the better. Xorg finally
set Ctrl+Alt+Backspace to be disabled by default about a year ago, to
prevent accidental data loss by newbies. Some people agree with this,
some people disagree, but the change is done. So it would have been
better to write "There is little hope, so can we please drop this
thread?" instead.
^ [1]. http://docs.sun.com/app/docs/doc/817-0708/6mgg6t7gv?a=view
Doug Barton
> On Tue, Dec 22, 2009 at 6:44 PM, Doug Barton <do...@freebsd.org> wrote:
>> There
>> is no hope whatsoever that you're going to get every Unix that has a
>> rational 'killall' command to change, so can we please drop this thread?
>
> It's never too late to change something for the better.
And yet there is ZERO interest in changing this in FreeBSD. Why don't
you start working on the various linux distros instead, and report
back here with your results.
Doug
--
Improve the effectiveness of your Internet presence with
a domain name makeover! http://SupersetSolutions.com/
_______________________________________________
Jason Spiro
> killall can be used by scripts which just works in the past, and will
> never notice the warnings.
On what scripts will nobody notice the warnings? For example, AFAIK, cron job
output is always mailed to root. The only scripts I can think of are scripts
called by web applications like PHP, and I can't think of any concrete case
where they would run killall.
> Also, killall is not "that" dangerous on
> FreeBSD, we should ONLY give warnings when it's really necessary,
> otherwise users would just ignore all warnings we gave to them.
>
> On the other hand, it seems to us that warning messages won't work, no
> matter how long we give it, it is being ignored by a majority of users.
Good points.
> Then users are already familiar with FreeBSD would have to learn what
> "fkill" is, and after all, having them to pay for mistakes made by
> commercial Unix vendors does not seem to be a fair option.
As I wrote elsewhere[1] in this thread, it seems to me the commercial vendors
made no mistakes here; only Linux and FreeBSD made mistakes.
> Well, I'd say it's too late for us to change since it's several years
> after we have 'killall' our way.
I replied to this in the last paragraph of text in [1].
> pkill have '-I', at least on FreeBSD...
There is no such option in pkill on Linux.[2]
^ [1]. http://thread.gmane.org/gmane.os.freebsd.devel.hackers/38308/focus=38332
^ [2]. http://linux.die.net/man/1/pkill
Jason A. Spiro
> And yet there is ZERO interest in changing this in FreeBSD.
As you can see elsewhere in this thread, I am discussing it with Xin.
So far, both he and the Linux killall maintainer have said "no", but I
am using rational arguments to try to convince each of them. I await
their next replies.
> Why don't
> you start working on the various linux distros instead, and report
> back here with your results.
All the Linux distros use the same killall, maintained by Craig Small.
I don't know when he got de-CC'ed from this thread. I have re-added
him to the CC field in this message.
And it seems to me that neither NetBSD or OpenBSD ship with a killall
utility, so there is no need to change such a thing as this in those
OSes.
Xin LI
> Xin LI<delphij<at> delphij.net> writes:
>
>> killall can be used by scripts which just works in the past, and will
>> never notice the warnings.
>
> On what scripts will nobody notice the warnings? For example, AFAIK, cron job
> output is always mailed to root. The only scripts I can think of are scripts
> called by web applications like PHP, and I can't think of any concrete case
> where they would run killall.
killall is used for instance, shutdown scripts. Yes you get the warning
message on your console but not the remote ssh.
[...]
>> Then users are already familiar with FreeBSD would have to learn what
>> "fkill" is, and after all, having them to pay for mistakes made by
>> commercial Unix vendors does not seem to be a fair option.
>
> As I wrote elsewhere[1] in this thread, it seems to me the commercial vendors
> made no mistakes here; only Linux and FreeBSD made mistakes.
I think we can hardly call it a 'mistake'. Having a command that do the
same thing what shutdown(8) should do doesn't seem to be the Unix way to
do things.
Speaking about commercial vendor, Mac OS X have the same killall as
FreeBSD have. Granted, Mac OS X is not something we consider as
traditional Unix, but it's certificated as Unix operating system after all.
>> Well, I'd say it's too late for us to change since it's several years
>> after we have 'killall' our way.
>
> I replied to this in the last paragraph of text in [1].
It's way too late to say something a "mistake" after about 15 years.
I think it might be reasonable to document the System V behavior and how
to do the same thing on FreeBSD in killall's manual page, but I'm afraid
that's all we can do nowadays, since FreeBSD users are already get used
with our killall behavior, changing the behavior/semantics after ten
years just make a mess, so please drop this.
>> pkill have '-I', at least on FreeBSD...
>
> There is no such option in pkill on Linux.[2]
Please talk with the authors of Linux pkill. In open source world a
well written patch would say more than a thousand of sayings.
Cheers,
--
Xin LI <del...@delphij.net> http://www.delphij.net/
FreeBSD - The Power to Serve! Live free or die
Doug Barton
> On Tue, Dec 22, 2009 at 7:00 PM, Doug Barton <do...@freebsd.org> wrote:
>
>> And yet there is ZERO interest in changing this in FreeBSD.
>
> As you can see elsewhere in this thread, I am discussing it with Xin.
> So far, both he and the Linux killall maintainer have said "no", but I
> am using rational arguments to try to convince each of them. I await
> their next replies.
Sounds like you have your bases covered, so there is no reason to
involve the list any further.
Doug
--
Improve the effectiveness of your Internet presence with
a domain name makeover! http://SupersetSolutions.com/
Daniel Tordable
root # killall *
System is going down INMEDIATELY
and then forget about the "shutdown" command to save some KiB of space?
Jason Spiro
> killall is used for instance, shutdown scripts. Yes you get the warning
> message on your console but not the remote ssh.
[snip]
> It's way too late to say something a "mistake" after about 15 years.
[snip]
> FreeBSD users are already get used
> with our killall behavior, changing the behavior/semantics after ten
> years just make a mess, so please drop this.
[snip]
> Please talk with the authors of Linux pkill. In open source world a
> well written patch would say more than a thousand of sayings.
Xin, I've already sent a message to the authors of Linux pkill suggesting that.
:)
Dear Xin and everyone else: Your messages, and the other messages that other
people have posted in this thread, are full of good points. I'm now convinced
that my original idea is not worthwhile. Thank you all for your replies.
Less than an hour ago, I've thought of another idea:
You could rename killall to something else, and update the manpage to show the
new name, then symlink or hard link "killall" to the new name forever. This
would encourage people to use the new name, but it would never force them to.
Xin, what do you think?
Phillip Spring
> Dear hackers, thanks for maintaining it on FreeBSD
you are welcome, and happy new year for everybody! :)
why rename killall? what's next? rename init? why not enhance kill? this is the end of alias? omg...
write shell script and name it, this is not a sin. it can be a crime... but don't worry, all freebsd users will go to hell anyways.
underdog.
my dream is a /lib/kill.so (and a job)
_________________________________________________________________
Fique protegido de ame�as utilizando o Novo Internet Explorer 8. Baixe j�, � gr�tis!
http://brasil.microsoft.com.br/IE8/mergulhe/?utm_source=MSN%3BHotmail&utm_medium=Tagline&utm_content=Tag1&utm_campaign=IE8_______________________________________________
Jason Spiro
[snip]
>
> You could rename killall to something else, and update the manpage to show the
> new name, then symlink or hard link "killall" to the new name forever. This
> would encourage people to use the new name, but it would never force them to.
>
> Xin, what do you think?
Since Xin and I CC'ed each other so often in this thread, Gmail automatically
added us to each others' Gtalk instant-messenger contact lists. Since it'd been
more than a month and Xin still hadn't replied to this message, I instant
messaged him the idea and asked him what he thought. He replied that, in his
opinion, this new idea is pointless: nobody would notice the renaming, and it
would just cause confusion. I asked, what if the tool would show a hint
onscreen if you called it using the old name? He said no. He said people can
just use pkill.
I hope Eitan Adler's patch[1], which adds a "verbose" option to pkill, will get
accepted, and that a similar patch will get accepted into the Linux pkill,
maintained by Albert Calahan.
^ [1]. http://thread.gmane.org/gmane.os.freebsd.devel.hackers/38636