Fredrik Eriksson schrieb:
> On Sun, 20 Sep 2009 18:11:27 +0200, clipka <ano### [at] anonymousorg> wrote:
>> You could generate a registry patch (I think .reg is the extension) 
>> that would be written to the HKCU tree, and somehow make sure that 
>> thing executed next time the user logs in.
>>
>> (The only thing which is a real nuisance there is that this doesn't 
>> allow you to /delete/ registry keys.)
> 
> Of course it does. What makes you think otherwise?

Experience.

Name any syntax in a .reg file that will make a registry key /disappear/ 
when that .reg file is executed.

(And I'm not talking about setting it to zero or an empty string or some 
such, I mean really /eliminating/ it.)

Post a reply to this message

clipka wrote:
> Indeed, a clear line cannot always be drawn; for instance, are POV-Ray 
> .ini files settings for POV-Ray, or are they part of the scene?

Generally, if the user can change them, they shouldn't be in the 
program-files directory. :-)

> Most software is pretty sane about this, 

Mainly because that's the only place users have permissions to write. :)

Altho I have had software that was licensed per-user that stored its 
licensing settings in the per-machine area. It was broken in many other ways 
too, and written by the CEO's son, so I wasn't surprised.

-- 
   Darren New, San Diego CA, USA (PST)
   I ordered stamps from Zazzle that read "Place Stamp Here".

Post a reply to this message

Orchid XP v8 wrote:
> remove this cache...

I've found that almost everyone stores junk with either the word "temp" in 
it or the word "cache" in it, so it's actually not too hard to do a scan and 
find most of the places accumulating cruft.

> The really fun thing is that you can't edit a user's registry chunk 
> without loading it first...

But not as the user.

-- 
   Darren New, San Diego CA, USA (PST)
   I ordered stamps from Zazzle that read "Place Stamp Here".

Post a reply to this message

clipka wrote:
> You could generate a registry patch (I think .reg is the extension) that 
> would be written to the HKCU tree, and somehow make sure that thing 
> executed next time the user logs in.

You have to load the hive. You don't have to load the hive under HKCU.

> (The only thing which is a real nuisance there is that this doesn't 
> allow you to /delete/ registry keys.)

Yes you can.

[HKEY_LOCAL_MACHINE\SOFTWARE\Blah\Blah\Blah]
"DeleteThisKey"=-

[-HKEY_LOCAL_MACHINE\SOFTWARE\Blah\Blah\Blah]

-- 
   Darren New, San Diego CA, USA (PST)
   I ordered stamps from Zazzle that read "Place Stamp Here".

Post a reply to this message

On Sun, 20 Sep 2009 18:59:07 +0200, clipka <ano### [at] anonymousorg> wrote:
> Experience.
>
> Name any syntax in a .reg file that will make a registry key /disappear/  
> when that .reg file is executed.
>
> (And I'm not talking about setting it to zero or an empty string or some  
> such, I mean really /eliminating/ it.)

http://support.microsoft.com/kb/310516

Scroll down to "Deleting Registry Keys and Values".



-- 
FE

Post a reply to this message

clipka wrote:
> BTW, does Linux provide such a feature as well? If so, where does the 
> "use count" live there?

Each package manager has its own way of dealing with such things. In the PM, 
it's somewhere in the mess of stuff stored under /var

I.e., it's handled by the package manager, not by install scripts and so on, 
so if you compile and install from source, it's not going to bump the use 
count on your libraries.

-- 
   Darren New, San Diego CA, USA (PST)
   I ordered stamps from Zazzle that read "Place Stamp Here".

Post a reply to this message

Orchid XP v8 wrote:
> Darren New wrote:
> 
>> Now I have a bat file that deletes all the crap I can find I run 
>> before backups.
> 
> I had that.
> 
> Then I discovered that it can't delete the IE web cache. (Due to some 
> kind of weirdness with it being system and hidden or something like 
> that.)

Well, you do have to turn that off. "attrib -h -s junkdir /s"

> So I switched to a Tcl script. 

That was the first one. Now I'm more lazy. :-)

> Even built a little Tk GUI for it, 
> so you can see how much longer it's going to take. (Deleting 0.8GB of 
> web cache takes a while...)

You can do this with "disk cleanup" tho.

> And then HQ IT decided that roaming profiles were too efficient, and 
> disabled them. Now every...single...time...I set up a new PC, I have a 
> stream of people asking me to come over and configure their email 
> settings. And people whining that they have to adjust their Word 
> settings manually for every PC they use. *sigh*

Sounds like you need to point out to your boss that machine inefficiencies 
are designed to reduce human inefficiencies.  Here's an example where you 
have people on your side - everyone actually working there hates the change, 
so you can tell them "I won't be able to get to fixing the problem for three 
hours. In the meantime, tell your boss you need roaming profiles."

Damn, you need a new job.

-- 
   Darren New, San Diego CA, USA (PST)
   I ordered stamps from Zazzle that read "Place Stamp Here".

Post a reply to this message

Fredrik Eriksson schrieb:
> http://support.microsoft.com/kb/310516
> 
> Scroll down to "Deleting Registry Keys and Values".

Ah, that's definitely news to me. So either they must have changed it 
since I looked last, or it wasn't documented back then. I must confess 
that it was quite a while ago.

Post a reply to this message

Darren New schrieb:
> 
> I.e., it's handled by the package manager, not by install scripts and so 
> on, so if you compile and install from source, it's not going to bump 
> the use count on your libraries.

That's what I wanted to know - thanks.

So, Windows is actually a lot better at this... /theoretically/.

Post a reply to this message

clipka wrote:
> nemesis schrieb:
>> Non-package-managed software installed by hand is of no concern to the 
>> package manager.  It will uninstall swiftly.
>>
>> I'm used to compile up-to-date libs required by some hand-compiled 
>> up-to-date software and install it to my ~/lib.  It's only annoying to 
>> do the usual "LD_LIBRARY_PATH=~/lib foo" everytime, but, hey!, that's 
>> why bash history is for. :)
> 
> I have no idea what you're talking about here, nor do I know whether you 
> understood my point.
> 
> What I mean is this: Suppose I install a package-managed piece of 
> software (let's name it "FooPNG" for now) that needs the libpng package. 
> Of couse the package manager will recognize, and ask me whether I want 
> to install the package as well. (Sure I do, yes please, go ahead.)
> 
> Now I compile this strange piece of software called POV-Ray, which 
> obviously isn't package managed. But it needs the libpng nonetheless. No 
> problem here, as the libpng is already there, so it compiles (and 
> installs) fine.
> 
> Now I decide to uninstall FooPNG.
> 
> Will the package manager uninstall libpng as well, figuring that only 
> FooPNG should rely on it? Will it leave libpng alone, figuring it 
> /might/ be used by /some/ other app? Or will it actually have any way of 
> /knowing/ that POV-Ray relies on it?
> 
> (I really don't care whether the package manager will uninstall 
> /swiftly/ or /slowly/, I care /what/ it uninstalls :-))

I understood your point.  Hand-compiled and installed povray or any 
other is of no concern to the package manager.  Thus, it'll uninstall 
libpng without caring if hand-compiled povray uses it or not.  It 
doesn't even know hand-compiled povray exists.  Nor should it.

In any case, povray 3.6 is featured as part of the Ubuntu multiverse 
repositories... :)

Post a reply to this message