Stephan Wiesand wrote:
> Hi Troy,
> 
> On Tue, 19 Sep 2006, Troy Dawson wrote:
> 
>> Stephan Wiesand wrote:
>>> Hi Troy,
>>>
>>> there were still a couple of minor problems. For example, it's broken 
>>> on 64bit. And I really didn't understand the [ ! -a <filename> ] tests.
>>>
>>
>> Thanks for the 64bit solution, I had forgotten about that.
>>
>> About the test
>>
>> [ ! -a $line/<filename> ]
>>
>> It is just checking to see if the file is there.  If there is a file 
>> there, I am assuming that we don't want to mess with it.  I also 
>> figure that if the file is there, then the directory is there.
> 
> But shouldn't that be "[ ! -e" instead? The above seems to always 
> evaluate to true. Maybe it's some black magic I don't get yet?
> 

Ah ... it's from my "unix in a nutshell" book, the test conditions page. 
  But now that I double check, it says the this is a Korn shell option. 
  But it is the check for "file exists".
Of course when I look through the man page, -e is for "file exists" and 
-a is for comparing two values.  Hmmm ... maybe when there isn't 
anything infront of the -a, it defaults back to checking if the file 
exists.  It works, but yes, I think sticking with the man page would be 
best.

>> [ -d "$line" -a ! -e "$line/%{cleanscript}" ]
>>
>> Although your solution is more thorough, it just seems redundant, 
>> doing two checks instead of one (unless I'm reading it wrong).
> 
> I agree it's a bit paranoid, but then this runs with root privileges, 
> and I wouldn't swear that the glob above will never ever return anything 
> weird, no matter under what circumstances. Also, I think
> [ ! -e "$line/%{cleanscript}" ] is not a guarantee that $line exists and 
> is a directory, is it?
> 

I think your right.  Let's go with your rpm as it stands.
I'm going to recompile and resign it, so that it's consistant with the 
same key and build machine as the rest of the errata.

Troy

> Cheers,
>     Stephan
> 
>>> I've put up a a slightly hacked version here:
>>>   http://www-zeuthen.desy.de/~wiesand/parentlock/
>>>
>>> Cheers,
>>>     Stephan
>>>
>>> On Mon, 18 Sep 2006, Troy Dawson wrote:
>>>
>>>> Hello,
>>>> This rpm is designed to fix the "Home Area in AFS" parentlock fix.
>>>>
>>>> The problem is triggered when your home area is in AFS, and you use 
>>>> firefox, thunderbird, and/or seamonkey.  Not always, but often, when 
>>>> you stop and restart firefox (or thunderbird, or seamonkey) the lock 
>>>> file, called parentlock, is not removed.  This makes it hard to 
>>>> start up your browser again.
>>>>
>>>> Ftp:
>>>> ftp://ftp.scientificlinux.org/linux/scientific/30rolling/testing/i386/RPMS/browsers/SL_firefox_parentlock_fix-1.0-3.noarch.rpm 
>>>> ftp://ftp.scientificlinux.org/linux/scientific/40rolling/testing/i386/RPMS/browsers/SL_firefox_parentlock_fix-1.0-3.noarch.rpm 
>>>>
>>>> Yum - S.L. 4.x
>>>>  yum --enablerepo=sl-testing install SL_firefox_parentlock_fix
>>>>
>>>> The rpm has already been tested on several computers, and we believe 
>>>> it works well.  But we would like others to test it more before it 
>>>> goes out.
>>>>
>>>> Thanks
>>>> Troy
>>>>
>>>>
>>>
>>
>>
>>
> 


-- 
__________________________________________________
Troy Dawson  [log in to unmask]  (630)840-6468
Fermilab  ComputingDivision/CSS  CSI Group
__________________________________________________