Thanks Urs, it looks like that works pretty good.
Now for the hardest part ... figuring out a name.
I'd like to put it all into one SL rpm, and just have the triggers be 
off the various packages.

I was thinking of calling it

SL_mozilla_afs_user

But I'm thinking the mozilla name might throw people off.  I thought of 
mozilla because all of these are mozilla based.

Troy

Urs Beyerle wrote:
> Hi Troy,
> 
> I agree with Stephan. I would like to see the fix in SL and I would like
> to have a separate rpm for the fix.
> 
> I use at PSI the following S01parentlock's for firefox, thunderbird,
> seamonkey (see below). Since four weeks no complains from our users so far.
> 
> What's different to your script:
> - a profile can have spaces within the profile name, therefore
> "$i/.parentlock" instead of $i/.parentlock
> - seamonkey has no profiles.ini, that's why I do a 'find -noleaf'
> instead of searching in profiles.ini.
> - to make it more simpler, I do also a 'find -noleaf' in case of
> thunderbird and firefox.
> 
> 
> $ more /usr/lib/firefox-1.5.0.5/init.d/S01parentlock
> #!/bin/sh
> #
> # removes .parentlock
> #
> 
> APP_FOLDER="$HOME/.mozilla/firefox"
> 
> find $APP_FOLDER -noleaf -type d -maxdepth 1 -mindepth 1 | while read i; do
>     if [ ! -L "$i/lock" -a -e "$i/.parentlock" ]; then
>         rm -f "$i/.parentlock"
>     fi
> done
> 
> $ more /usr/lib/thunderbird-1.5.0.5/init.d/S01parentlock
> #!/bin/sh
> #
> # removes .parentlock
> #
> 
> APP_FOLDER="$HOME/.thunderbird"
> 
> find $APP_FOLDER -noleaf -type d -maxdepth 1 -mindepth 1 | while read i; do
>     if [ ! -L "$i/lock" -a -e "$i/.parentlock" ]; then
>         rm -f "$i/.parentlock"
>     fi
> done
> 
> $ more /usr/lib/mozilla-seamonkey-1.0.3/init.d/S01parentlock
> #!/bin/sh
> #
> # removes .parentlock
> #
> 
> APP_FOLDER="$HOME/.mozilla"
> 
> find $APP_FOLDER -noleaf -type d -maxdepth 2 -mindepth 2 | grep -v
> firefox | grep -v
> Cache | while read i; do
>     if [ ! -L "$i/lock" -a -e "$i/.parentlock" ]; then
>         rm -f "$i/.parentlock"
>     fi
> done
> 
> 
> Cheers,
> 
>     Urs
> 
> 
> 
> 
> Troy Dawson wrote:
>> Stephan Wiesand wrote:
>>> Hi Troy,
>>>
>>> On Fri, 15 Sep 2006, Troy Dawson wrote:
>>>
>>>> Hello,
>>>> I know the AFS bug with the new browsers is really a 'big lab'
>>>> problem (any
>>>> lab big enough to use AFS for home space), but I was wondering what
>>>> everyone
>>>> thought if I put the AFS fix into the generic S.L. mozilla based
>>>> packages
>>>> (firefox, thunderbird, and seamonkey)?
>>>>
>>>> Do you think I should do that (generic S.L.)?  Or just patch Fermi's?
>>> We would certainly welcome having the fix in the SL packages. And if
>>> these are the only choices, from our point of view: please go ahead
>>> and do it.
>>>
>>> However, I think there's a third choice, cleaner and in line with SL
>>> traditions: put the fix into a tweek rpm. Maybe something like what
>>> the "official" flash-plugin rpm is doing:
>>>
>>>  - trigger on installations of firefox, thunderbird, seamonkey
>>>   - search /usr/lib{,64}/ for their install dirs
>>>   - copy over the fix script
>>>
>>> This would make the fix optional, retaining full binary compatibility
>>> with
>>> upstream for those who don't need the fix. And it's probably less work
>>> to maintain in the long run, since we'd have a single noarch rpm for all
>>> architectures and all [future] releases of SL, firefox, thunderbird,
>>> and seamonkey.
>>>
>>>> Also, from what I read, the fix is to put a script in
>>>> /usr/lib/<browser name and release>/init.d
>>>>
>>>> That looks like
>>>>
>>>> #!/bin/sh
>>>>
>>>> [ -e  $HOME/.mozilla/firefox/profiles.ini ] || exit 0
>>>>
>>>> for i in $(grep Path= $HOME/.mozilla/firefox/profiles.ini | sed
>>>> s/Path=//); do
>>>>     [ ! -L $HOME/.mozilla/firefox/$i/lock -a -e
>>>> $HOME/.mozilla/firefox/$i/.parentlock ] &&
>>>>       rm -f $HOME/.mozilla/firefox/$i/.parentlock
>>>> done
>>> Yes, this has worked well here since we implemented it. It solves the
>>> problem, and I'm not aware of any ill effects. I don't remember exactly,
>>> but I think the filename has to start with "S", and the file has to
>>> be executable or end in ".sh". Anyway, our fix here uses
>>> "S01parentlock" and is mode 0755, and that works.
>>>
>>> Cheers,
>>>     Stephan
>>>
>> Hi Stephan,
>> You are correct, this really should be a SL rpm.  I've been patching
>> rpm's for fermilab so much the past two week's I hadn't even thought
>> of it.
>>
>> I'll have those SL rpm's done soon, hopefully.
>>
>> Oh, on a related note, we need a fix for this for Seamonkey.  I'm
>> having a hard time finding the user Path for seamonkey.  Does anyone
>> have a script like this that works with seamonkey?
>>
>> Troy


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