On 10/15/2011 2:08, andrel wrote:
 > That said, windows does have a priority mechanism, and the taskmanager is
 > running at a high level, two levels higher than Blender and Matlab, so why
 > does it not work?

Because you're confusing disk priorities with CPU priorities, and you can't 
easily interrupt a disk seek. That's kind of what I've been saying. In the 
length of time it takes the Task Manager to get to the next page-in, Blender 
has already run far enough to trigger a page fault on its own.

 > If it is because someone is using it for other things, like POVray, Blender
 > or a game fire that person. If it is because of tasks that should run there
 > fire the IT manager or the financial director, depending on who was
 > responsible for buying the wrong infrastructure.

But you might not know. And you might not be able to easily tell. And you 
might be running multiple tasks on the same machine, or even multiple 
operating systems nowadays.

 > Windows is open source now? I missed that.

No. Fix it in Linux. :-)  Linux does exactly the same thing.

Post a reply to this message

On Mon, 10 Oct 2011 11:11:45 +0100, Invisible wrote:

>>> So, just because it does strong authentication, you think that means
>>> the actual data is encrypted?
>>
>> It's actually a certificate verification message, not a 'strong
>> authentication' message.  It's asking about an SSL certificate that's
>> used to encrypt the entire communications channel.
>>
>> You know, like actual security.
> 
> Fact: It doesn't matter how strong the authentication process is. This
> does not automatically mean that the data that follows is encrypted in
> any way at all.

No, it doesn't - the fact that it actually *is* encrypted is what means 
that it is, you know, encrypted.

>> Don't believe me?  Fine, I'll do a wireshark trace on it.
>>
>> Nope, 1200 packets, nothing in the clear.
> 
> And how do you tell whether random binary data is encrypted or not?

There's nothing "in the clear".  I connected to the system, opened a CMD 
window, and listed directory contents.

That, plus the fact that it, you know, actually is *documented* to be 
encrypted.

The fact that you don't *believe* the documentation doesn't, you know, 
actually mean it's not encrypted.

How do you know your VPN is encrypted?  Because it *says* so?  What are 
you, nuts? ;)

>> "128-bit encryption, using the RC4 encryption algorithm, as of Version
>> 6.
> 
> RC4? Man, how ancient is that? You realise it was a weakness in RC4 that
> allowed WEP to be broken, right?

Yes, I do.  However, *weak* encryption is still, you know, *encryption*.

>> Nope, I guess you're right.  Adding 128-bit encryption isn't security.
> 
> Fact: The number of bits in the encryption key is not directly related
> to how secure the encryption is. Triple DES has a 168-bit key, and it's
> widely considered far too insecure to use.

But it's still encryption.  You asserted that it's not encrypted.  I 
proved that it was.  Now, if you want to talk about encryption 
*strength*, that's different than, you know, whether it's encrypted or 
not.

>> "Support for Transport Layer Security (TLS) 1.0 on both server and
>> client ends (set as default)."
> 
> Now that's more like it.
> 
> (Sadly, on further investigation, it appears that TLS 1 still uses RC4
> or Triple-DES. So much for HTTPS being secure...)

But it's still *encrypted*, which you categorically claimed it wasn't.

>> Clearly I don't have a clue what I'm talking about.
> 
> I'm still left wondering how many of these features are actually turned
> on by default. Every Windows protocol I know of sends everything
> unencrypted by default, and most of them offer no possibility of adding
> encryption. I'd be rather surprised if RDP is different.

Well, it's just documented as being enabled by default.  Like your VPN.  
How do you know your VPN is actually encrypted?

>> Oh, and I pointed you at an SSH server for Windows.  It comes with
>> Cygwin.
> 
> Right. I didn't know about that when I set this up.

You knew about it before you made this post.  So now you know it's 
available so you can use it.  That is if you believe that SSH is actually 
encrypted.  After all, you just have the documentation to tell you that - 
because it's a stream of random binary data.  It may well not be 
encrypted.  You don't know.  Just like RDP.

Jim

Post a reply to this message

On Tue, 11 Oct 2011 09:00:11 +0100, Invisible wrote:

>>> Also, it contains far less information than a STOP message does.
>>> Basically
>>> an error code (plus textual translation) and a memory address. Good
>>> luck working out, say, what was *at* that address or anything...
>>
>> That's why the power light flashes 3 times before the message comes up.
>> That's you're cue to hit DEL on the serial terminal plugged in to the
>> back to launch the kernel debugger.
> 
> Uh, what's a serial terminal? Also, what's a kernel debugger?

*facepalm*

Jim

Post a reply to this message

On Thu, 13 Oct 2011 09:20:00 +0100, Invisible wrote:

> On 12/10/2011 05:17 PM, Darren New wrote:
>> On 10/12/2011 1:13, Invisible wrote:
>>> Now given that the Amiga is a home computer, how many homes have a
>>> VT100 just laying around?
>>
>> The VT100 is not the only serial terminal. Anything that talks RS232
>> serial is sufficient. You know, like, another Amiga, say?
> 
> Oh, OK.
> 
> So what can you actually do with a "kernel debugger", anyway? Just look
> around the machine's memory or something?

Yes, that's exactly what a kernel debugger is for.  Debugging the kernel, 
hence the name. ;)

Jim

Post a reply to this message

On Mon, 10 Oct 2011 11:00:02 +0100, Invisible wrote:

> On 09/10/2011 10:00 PM, Jim Henderson wrote:
> 
>> The point was that Andy said that binary blob data could be stored in
>> the registry and not a Linux configuration file.  Point is, it could be
>> stored in a Linux data file, but since it's common to change
>> configuration items with an editor in Linux, it's not common to use a
>> binary format (though the timezone file is an exception to that IIRC).
> 
> The point being, if you want to store some binary data in the middle of
> a textual configuration file, you have to base64 encode it or something
> (which is less efficient). If you want to stick some binary data in the
> registry, you can just store it as binary.

No, you don't have to encode it as base64.  It's perfectly legal to 
create a file with text components and binary components.  Because it's 
just a file.

Jim

Post a reply to this message

On Mon, 10 Oct 2011 10:00:49 -0700, Darren New wrote:

> On 10/9/2011 21:43, Jim Henderson wrote:
>>> I didn't catch that assertion. OK. It's fairly unique in being a
>>> standard searchable format with an API, but not in that it stores
>>> binary data.
>>
>> grep -ri text /etc/*
>>
>> Works pretty well for being a standard searchable format on Linux. ;)
> 
> Without an API. Or transactions. And there's really no standard
> searchable format there besides "a text file". Tell me what standard
> Linux tool will give you a list of all the Document roots that Apache is
> serving.

grep.

Jim

Post a reply to this message

On Mon, 10 Oct 2011 10:42:46 +0100, Invisible wrote:

>>>> Well, yes and no.  Users of SUSE products (openSUSE and SLE*) often
>>>> do know how to do the manual edits, but prefer using YaST anyways.
>>>
>>> If you pull up the documentation for (say) Apache, it won't tell you
>>> how to use the Apache YaST module. It will tell you how to edit the
>>> underlying text file.
>>
>> If you use openSUSE for configuration, you use the openSUSE
>> documentation to see how to use YaST to make those configuration
>> changes.
> 
> Wait - YaST has documentation?

Um, yes.  man yast for starters.

>> And if that
>> doesn't get you where you need to be, you ask a question in the
>> community.
>>
>> OSS is big about community.
> 
> That sounds very nice and all, but if I'm trying to quickly set
> something up, I don't really want to have to go onto the Internet and
> beg for help, and then spend a week or two hoping that somebody
> knowledgeable will just happen across my message and actually take the
> time to give me a helpful response. I want to read a manual that tells
> me how to do this RIGHT NOW.

Right, you'd rather struggle with it for weeks and weeks and then bitch 
about how difficult it is to do anything.

Instead of asking a question and getting an answer within a couple of 
days so you can actually use it.

>> I guess I imagined all those Technet articles that have the warning I
>> sited earlier about how editing the registry can screw your system up.
>> That must be it, because of course Microsoft would *never* recommend
>> you do something that might bork your system.
> 
> The warning is partly there because if you're a clueless user who
> doesn't know how to work a computer properly, it's very easy to do a
> hell of a lot of damage using a registry editor. Personally, I have
> almost never borked my system by editing the registry. The only time
> it's happened is when I started deleting stuff at random in a desperate
> attempt to make something work. If you follow the instructions, it works
> fine.

So, modifying it is dangerous.  Or not.  The warnings don't exist.  Or 
they do.

Gotcha.

> My point remains: It's very uncommon to /need/ to touch the registry in
> the first place. Whereas under Unix, the text configuration files are
> the first port of call, not the last. That's just the difference in
> design mentality.

Unless you use YaST, webmin, or one of a myriad of other Linux 
configuration tools.

>> Obviously you don't know many Linux users.  I know at least 5,000, and
>> many of them not only love and use the GUI, but tend to have religious
>> wars over which GUI is better.
> 
> And yet, the vast majority of all Linux software is strictly CLI-only,
> and developers always seem to expect somebody /else/ to build the pretty
> front-end for it.

That's just incorrect.  But since you believe it is, it must be true, 
regardless of evidence to the contrary.

>>> I notice that there's always a lot of stuff "happening" with Linux.
>>> I'm never sure what the hell any of it actually /does/.
>>
>> Linux (and most OSS software) evolves rather than going through
>> discrete cycles.
> 
> Sure. The "release early and often" approach. I'm just saying, as an
> outsider, it's not always clear what actually changed between versions.
> (I guess often it's stuff under the hood that you won't notice anyway.)

Of course it isn't, that's what evolutionary development models do.

>>>> Transactionality is a function of the filesystem, and I use a
>>>> journaled filesystem.
>>>
>>> Doesn't stop two scripts both trying to update the same config file at
>>> the same time. If you do that with the registry, it works. Because
>>> it's a proper database engine, not just a flat file.
>>
>> I'm not sure how "proper" that database engine is - IIRC, it's JET, and
>> most DBAs that I know would say that's certainly not a proper database
>> engine.
> 
> JET is no match for an enterprise database engine, sure. But it's more
> transactional than a flat file.
> 
> Also, I'm not completely sure that the registry is actually JET. It
> might be, but I didn't think it was. For one thing, registry files grow
> as needed, but never shrink. I don't think JET has that limitation.

FFS, *Active Directory* is (was) JET.  Maybe they moved to something else 
now, but I know from personal discussions with AD architects at Microsoft 
that it is/was JET.

JET has been MS' solution for simple database storage for a number of 
years, possibly decades.

>>> Personally, I'm not very impressed by the Windows Update system. Like,
>>> it'll install a bazillion updates for IE6 in the same session as it
>>> also installs IE8. And then you go back and it wants to install a
>>> bunch of IE8 updates. Um, why couldn't you do that the first time
>>> around??
>>
>> Yep, I've been frustrated by that as well.
> 
> I'd ask if Linux gets this right - expect Linux generally won't replace
> one version of an application with a totally different one just because
> you asked for security updates...

Depends on the distribution and the updater, but on openSUSE, that's true.

>>> I love how multiple courts have proved that what MS is doing is
>>> illegal, and as a result they have received NO PUNISHMENT OF ANY KIND.
>>> That's such a big motivation for them to stop casually disregarding
>>> the law...
>>
>> Oh, I don't know, having to admit that Firefox is a reasonable browser
>> to use and they should change Windows architecturally to decouple IE
>> from it (or at least loosen the coupling) is a pretty significant
>> sanction.
> 
> Admitting you're wrong is one thing. But they did something illegal;
> where is the *financial* pain for that?

Obviously you missed the fact that they paid fines to the EC for their 
illegal activities.  And they had to reengineer some things - and 
engineers at commercial software companies don't tend to work for free.

>> Maintaining multiple versions of an entire operating system can be time
>> and resource intensive.
> 
> Maybe it's there then. Still seems like a fairly tiny price for, you
> know, BREAKING THE LAW...

Go look at the fines they paid.

Jim

Post a reply to this message

On Mon, 10 Oct 2011 10:52:52 +0100, Invisible wrote:

>>> Oh, tremendously. But they're still a PITA compared to Windows'
>>> explorer, methinks.
>>
>> Depends entirely on what you're used to.
> 
> Not /entirely/, no. Many things are subjective, but not all of them.

When it comes to user selection of tools, yes, they are.  If I find 
something easy to use, who are you to tell me "no, that's not easy"?

>> In VMware, if you "pause" the VM, VMware grabs the state and commits it
>> to disk, freeing up the memory for other VMs.  I find VMware's usage
>> makes a lot more sense.
> 
> Personally, I'd prefer having two separate buttons for these functions.
> Sometimes I just want to pause a VM just for a sec so I can use the CPU
> power for something else. So I hit pause in VMware, and facepalm as I
> have to wait ten minutes for it to trash the hell out of my HD. The
> /entire/ computer is unusable until this finishes. And when I decide to
> unpause the VM, we go through the same dance again.

Yes, two separate functions would be good, I agree.

> Also fun: Pausing a VM makes the display go away. So it's useless if
> you're trying to quickly pause the display so you can *read* that error
> message that flashes past too fast to see...

I could see that being frustrating, having been there myself.

Jim

Post a reply to this message

On Mon, 10 Oct 2011 10:44:54 +0100, Invisible wrote:

>> A more appropriate comparison is latest against latest.
> 
> I don't know how big the latest version of Windows is. I've never
> installed it. More to the point, I don't know of anybody who's using it
> yet.

Windows 7.  You might have heard of it.

And I've got it installed in a VM, as I'd mentioned before.

>>> It's still frustrating that I have to download multiple GB of data for
>>> libraries that I'm not actually going to use, just because the package
>>> dependency system isn't a little more fine-grained.
>>
>> You don't know that you're not going to use them.
> 
> That's an absurd argument.
> 
> If the PC doesn't have a sound card then I know, for a fact, that I
> don't need the sound daemon. If there are no Windows machines on the
> network then I know, for a fact, that I don't need Samba. I could go on.

Are you never ever ever ever going to install a sound card?  Put your 
machine on a Windows network?

It's a weak argument, I'll concede; it's not an absurd argument.

An absurd argument is saying that RDP isn't encrypted when every piece of 
documentation about it says that it is. ;)

>> If you want something more specialized, you can do a custom
>> roll-your-own with SUSE Studio.  No need to download anything you don't
>> want, because you build the distribution on a remote system using a web
>> browser, test it, and then download it.
> 
> Now that's what I'm talking about... That sounds actually useful.

Indeed it is.

>>> Yes. Because spending a week trying to fix something it *totally* the
>>> same as just giving up and complaining about it. Oh, wait...
>>
>> You spent a week trying to fix something that had you asked a question
>> in an online forum, you probably could've gotten an answer for in a day
>> or two.
>>
>> And then you complained that you couldn't figure it out.
>>
>> Yeah, ultimately, you gave up without asking for help.  If you'd asked
>> for help, someone probably would've been able to help you.
> 
> 1. You're assuming that it's /possible/ to fix this.

I'm basing that assumption on years of experience in troubleshooting 
Linux systems.

You're giving up and assuming it's impossible to fix it, so why bother 
even asking for help?

*That's* absurd.

> 2. You're assuming that had I asked, somebody would have actually
> bothered to reply. And that their reply would have been helpful.

Again, based on decades of experience in online forums.  Is there a 
chance of a useless answer or no answer?  Sure.  But I guarantee you you 
won't get an answer IF YOU DON'T FUCKING ASK.

>> You don't need a super expert to explain it.  You need someone with
>> more knowledge than you have to explain it.  Plenty of normal everyday
>> non- technical end users manage to use it without problems.
> 
> As I say, I can get Linux to /work/ OK. Indeed, my dad still uses it on
> a daily basis. It's just that one or two things - like getting the
> package manager to install just the packages I actually need - are
> annoyingly fiddly.

If you asked for help in understanding it, you might just learn something.

Heaven forbid *that* should happen. ;)

Jim

Post a reply to this message

On Mon, 10 Oct 2011 09:55:08 -0700, Darren New wrote:

> On 10/9/2011 21:42, Jim Henderson wrote:
>  > Are you sure about that?
> 
> Pretty sure. Certainly people like suing Microsoft more than they like
> suing open source projects. :-)

You haven't been following the legal battles against Android and the 
extortion Microsoft has engaged in over supposed 'patent infringements' 
Linux has on Microsoft patents.

Something Microsoft rattles a lot of sabers over, but which they refuse 
to be specifically precise about *which* patents are involved, because OSS 
developers would work around that.

>> If Acrobat Reader crashed a Windows box, it would be Adobe's problem
>> regardless of if Microsoft distributed the file or if they got it from
>> Adobe.
> 
> Yet, oddly, people blame Windows when a video driver crashes. Fancy
> that.

Odd that nobody's ever sued Microsoft over that (at least not that I've 
heard of), but somehow there's a threat of that happening with Acrobat 
Reader, apparently.

>> But a "repository" for Windows is just a "download" site.  It doesn't
>> include actual software management elements per se.
> 
> The management elements are built into the software you download. That's
> why, for example, you have things like "Programs\Common Files" and
> "Add/Remove Programs" and things like that.
> 
> Now, yes, the fact that each program manages its own means you have to
> deal with legacy software repositories, so it doesn't always work as
> cleanly as it might. DOOM still installs to C:\DOOM or some such.
> However, there are actual software management elements available, *if*
> you want to make use of them. Many small projects don't.

Installation directories aren't "software management elements" - things 
like dependency resolution are.

Jim

Post a reply to this message