Web has its own problems.
on
The Birth of vi
·
· Score: 2, Insightful
You make the point well enough about all the security. That means it's a lot saner to run your web configuration through Apache or something. It also means you'll have to deal with issues like a username/password, and you'll probably want to proxy the whole thing through SSH or something (or you'll need an SSL-enabled Apache)...
Unless you're doing something AJAX-y, changes do require you to fill out a form and hit "submit", and wait for a response. I don't really see how this is different than editing a config file and pinging the app. But if you're going to do something AJAX-y, we're getting pretty heavy into the realm of external dependencies -- this is no longer something you want embedded into your app, the way CUPS configuration is.
And I don't really want to have to install Apache, mod_ssl, and some AJAX libs just to configure, say, an IRC server, or a DNS server.
Besides: It doesn't have to require pinging the app. I seem to remember exim checking the timestamp on its config file with every mail received, so if you made a change, you could expect it to take effect instantly. Another way to do this would be inotify -- I believe vim writes all your changes to a temporary file, then 'rename's the temporary file on top of the original one. Thus, as soon as you see a new file there, you can probably parse it -- worst case, if you don't trust the text editor, you can wait a second or two for the inotify events to settle.
Now, there are advantages -- you can actually display a finite set of options, you can show the user what their system can actually handle, and there aren't going to be any syntax errors. But none of these have ever really been a problem for me -- I can crack open just about any config file, read the comments, and make a few changes, and be confident that it'll work.
I think this is pretty much how it's going to be from now on.
Major innovation will be on Linux and OS X. It will take a month or two for Linux to absorb new OS X features, and OS X will generally absorb some of the good ideas from Linux in the next release of their OS, which seems like it's going to be every couple of years. Linux will always have more features than anything else, and OS X will always be perceived as "easier" in some way.
Windows will always be just slightly behind the last version of OS X. That is, Vista and Leapord come out, and Vista will be almost as good as Tiger, with Leapord completely blowing it away. And I'll take the features I like from both, and implement them in Linux, and still have the features neither will touch, like a real package manager.
In any case, I think it's more compelling to make that kind of comparison. For instance, if we were able to show that Ubuntu Dapper is better than XP, that'd be a very compelling argument, because we could then say "Oh, by the way, we're on Edgy now, but XP can't even begin to compete with that."
Yes, it was the MM10. I didn't trust Sharp again, though, for a couple of major reasons:
Sharp tech support refused to replace my hard drive when it died the first time, no matter how many Linux log messages I read to them, until I wiped it and reinstalled XP on it. It occurs to me that they may have flat-out refused if I'd attempted to get a Windows refund -- which, in hindsight, I should've done.
I had to send it back to them several times, and it came back with fewer screws each time.
It was damned thin -- it apparently used the same kind of hard drive you could get in an iPod. So, the second time it failed, I figured I'd just buy a new hard drive and make it usable. Unfortunately, the only compatible hard drives I found were sold out, and Sharp refused to tell me where I could buy another hard drive. The only way to replace it is to either try an iPod hard drive (and hope it has the right interface), or pay Sharp some $100-200 service fee just to look at the thing and tell me if they can replace the hard drive. That's right -- they won't sell me a new hard drive, but they'll happily charge me hundreds of dollars to decide if they want to install a new one themself -- plus the cost of that new hard drive, if they deign to replace it.
No hard drive? Fine, it's got built-in WiFi, and it'll boot off USB. I should be able to plug in a USB key, boot it to a minimal Linux initrd, and have that Linux open a VPN connection to a fileserver and operate as a diskless, wireless machine. Unfortunately, even that doesn't work, as it used to take around 20 minutes to get past the splash screen when a hard drive wasn't connected (or working)... But, when I tried it again a month later, it never got past that screen. The machine is officially dead now.
The thing was just too flimsy. I can drop my Powerbook and it's fine, drop this and the hard drive's gone, or the screen cracks, or something.
Now, I've discovered how much I dislike certain aspects of the Powerbook, and I could be talked into getting another Actius, or something like it. But this time around, I want it to be able to play DVD video -- the MM10 choked on that, either the CPU or the video card, but it was not just Linux -- and I want it to have at least 40 or 50 gigs of storage, unless it's all solid state. Other than that, it's software (Linux) issues that I can fix myself...
No business person in their right mind would rewrite all their COBOL code into C or Java just for the sake of modernization. That would be foolish and stupid, and they would deserve to be fired from their jobs.
That's making a big assumption -- that this COBOL code all works, and that no one needs to change it much. That may be true for a lot of things, but as soon as you actually have to maintain the fucker, it starts actually looking cheaper to rewrite it -- even in Java -- than to try to tack on the features you need to the COBOL monstrosity.
My Powerbook -- well, the screen is dead for the moment, but it has been a router killer for some time now. I haven't been able to figure out whether it's OS X, the VPN, the SSH, or what, but everywhere I go with this thing, routers die and have to be reset (pull the plug). Sometimes it doesn't happen for days, sometimes it happens every hour or so, sometimes I open the thing up from sleep, get all connected to the wireless, and watch it kill the router.
I'm hoping that Vista will convince these router manufacturers to get their act together. Even if it is somehow OS X's fault, or my fault, I shouldn't be able to DOS a router that easily.
And if you're going to setup a dedicated Samba server, you really should make it a full domain controller. There's little point in doing P2P filesharing at that point.
Except that it's easy to set up (there's even a GUI for that now, I think), and it's about what they're used to. If I was going to really change things around there, it would be neither Samba nor domain controllers, it'd be a Linux fileserver, a ton of Linux desktops, and a single Windows application server. But that would be a huge project, and I can't make a good business case for it -- and there isn't too much short of that to hold my interest.
Well, I'd say in your analogy teaching users how to administrate a domain is more like teaching them how to solder a new pipe.
True enough. In this case, though, all they need to do is be able to setup a new box (and account), and probably tear down the old account and permissions. According to you, that's actually pretty easy -- so we give them a printed list of step-by-step instructions.
Ideally, of course, I'd have a script to handle this for them. Input the name and mac/ip address of the new box, then a username and password, then: done. But it really does sound like that's about what you do anyway.
Let's see -- first off, it would help if Microsoft had actually spelled out how this is supposed to work, rather than requiring us to have a copy of every word processor since WordPerfect and carefully reverse-engineer its behavior in every case.
Second, you're right, old format -> OOXML -> old format... yeah, that would kind of suck. However, I don't get why you would ever want such a loop to work. Old format -> OOXML -> some other format -- now that makes sense. Think of it this way -- you can burn your mp3s as music CDs, then rip those CDs into mp3s, but you're going to lose something.
But keep in mind: If OOXML -> old format is to work, it must be able to cover anything OOXML might do, including hacks to make the old format look right.
And more to the point, have you actually looked at these? Some of them seem entirely too simple: Emulate WordPerfect 5.x line spacing. Really, how hard would it be to simply have a generic <p spacing="0.8"> (just an example, I don't know how this is done in practice or exactly what the difference is) in your conversion of WordPerfect to OOXML? And, going the other way, couldn't you just perform a similar calculation to determine line spacing?
I mean, you're right, it takes a bit of effort. Yeah, your conversion tool takes a little more effort to write than it otherwise would. I think that's an acceptable loss, considering it saves a massive amount of effort from every fucking word processor to implement this standard, EVER.
As for some of this other stuff, some of it seems inevitable, if you want the document to look perfect. And if you accept that, then yes, you need these tags -- but would it've killed Microsoft, in their 6000 word "spec", to spell out what an app's actually supposed to do? Specs are supposed to be the basis for implementations -- they should not have to refer you to another implementation -- and it certainly doesn't help Microsoft's case of "open"-ness when it would cost me a significant amount of money to obtain all the word processors they refer to.
I, for one, don't accept that. The vast majority of these tags seem to be referring to specific weird behavior, like character spacing being slightly off, when they could've simply given us an option to specify how much space between each character. That is, whether it's for "backwards compatibility" or not, you should be able to specify... oh, let's say:
Now, it's true, the second one takes more space. That's why we use compression techniques -- both actual zip compression and programmatic compression. For instance, we could do something like this:
<shapedef id="CheeseWedge">
<point x="0" y="0"/>
<!-- you get the idea... --> </shape>
<!-- now, somewhere in the document: --> <shape id="CheeseWedge"/>
Worst case, you can use XML entities.
Now, you're right, half-baked conversion tools will screw you over -- but so will half-baked software. And really, after any conversion like this, doesn't it make sense to go back and check that the document was converted properly before deleting the original file?
The point of a new standard is not to support tons of legacy documents and software. It's to make sure we're all on the same page moving forward, and Microsoft's 6000 page monstrosity doesn't help much with that, especially when it's not even complete till you buy and reverse engineer 16 years worth of word processors.
Please give us direct access to the device -- none of this CompactFlash HD emulating crap. Windows can emulate a hard drive if it wants to -- Linux is developing an FS specifically designed for Flash devices, that operates directly on the Flash, not even as a block device.
Remember -- Flash is funny. You don't overwrite, you erase, then write -- in fairly large blocks. You want to intentionally fragment stuff, so you write evenly over the memory, so you don't wear it out as quickly -- but you also want to keep the chunks defragmented, even delay writes a little (Reiser4/XFS style), so you can overwrite a whole block at a time.
All in all, I'd much rather trust Linux/jffs2 to manage my Flash than some arbitrary hardware inside a CompactFlash device.
And speaking of OS choices... Why are you so desperate for OSX on such a device? I mean, you realize you've just reduced yourself to grovelling because your OS vendor needlessly restricts the hardware you run it on? You realize that the moment someone produces such a device, I'll be able to run Linux on it, but unless you're very lucky, you won't be able to run OS X without cracking it?
Regarding battery life -- my old Sharp Actius had some 9-10 of battery life, and it was a subnotebook. If you want battery life, it's simple: Sacrifice performance, and don't do much with it (read a book or something). Unfortunately, it doesn't seem like any manufacturer wants to do that.
If I remember right, PS3 Linux (or any other OS) runs in a hypervisor. That means virtualization -- it means that they actually can prevent Linux from gaining access to the hardware, just as I can run Windows safely inside qemu on my Linux and deny it access to my hardware.
So, it's not just that there are no drivers, and that you can hope someday someone will reverse-engineer it enough -- no. They'd have to crack it and reverse-engineer it -- basically like any kind of Xbox Linux.
As far as whether HD video works, I haven't gotten a straight answer from anyone. We know 3D acceleration doesn't work, and we know you get a framebuffer. What I don't know is whether the framebuffer is accelerated enough to play HD video or 2D games. But keep in mind -- even if it can play HD video, it seems doubtful that you'd actually be able to play a Blu-Ray disc.
You take a mishmash of code and build a polished product that anyone can use with very little effort and no technical ability.
No, actually, they wrote new software that runs on Linux, as I recall. Nothing to stop you from releasing proprietary software which runs on Linux. But, they do prevent us from actually using the code they did modify -- and by the way, they do give us the code, they just don't let us use it.
But if I take some unsupported code (Unsupported in terms of if it doesnt work, then I have no way of getting you to fix it in a reasonable timeframe) and turn it into a working, supported product why should I not be able to then protect my work from plagurism.
Nothing in GPL, either 2 or 3, prevents you from protecting it from plagarism. But I don't think "plagarism" means what you think it means.
Anyone who has had to work in the software development field professionally will probably agree that writing a product is the easy bit (and fun too). The shit bit is having to help users who think the manual is something YOU should read then tell them what is says.
So what? What does this have to do with anything?
The reality is that open source software would still be a hobbiest only affair without the freedom that the GPLv2 grants to make professional, user friendly products and then protect that product from being ripped off by a competitor who can produce it cheaper as you have done all the polishing.
First off -- TiVo is a piece of hardware. Second, others have indeed done TiVo clones, some with Linux, some without.
But more relevantly, there is nothing in GPLv3 which prevents you from making a professional, user friendly product. You certainly can -- and it will be all the more user friendly for me if I can actually modify your code.
Besides, if you truly believe open source would be a hobbyist affair without tivoization, you should look up Ubuntu, RedHat Enterprise Linux, and the vast numbers of Debian servers around the world. Linux doesn't need TiVo, and it doesn't need you. After all, what's the point in letting TiVo use it, if we get nothing back?
One more thing to remember: You're getting the code for free. You do not get to bitch about us changing the terms. Nothing is stopping you from developing your own TiVo DRM monstrosity, but you will not be doing it with my code -- at least, not for free.
Hmm. Looks about right. If I read that link right, it's comparing PPC-on-Rosetta-on-x86 to native x86.
My point was, Rosetta runs slower than native, of course. But, the x86 processor is itself much faster than a PPC processor. Not because of the architecture -- because of raw speed. Thus, even if you're running entirely PPC apps, a MacBook can still feel like an upgrade -- and of course, everyone recompiles for x86 (or "universal") quickly enough, because the x86 ones are that much faster.
What we'd want to avoid is feeling an actual loss in speed when switching to the new arch. In order to beat Worse Is Better, you have to be much better, enough that it's really a compelling switch. That's why x86_64 works -- it is a lot better (additional registers, more RAM), not much more expensive, and has a perforamnce hit of exactly zero for 32-bit code.
Yes, you're absolutely right -- and if you read my post carefully, you'll find that's roughly what I said. Except for the part about it being 6 books... that's a technicality, anyway.
You're a computer tech and you don't want to learn anything new?
On the contrary, I learn new things all the time -- just things I like to think of as useful. I don't want to learn Active Directory, because I'd much rather set people up with Linux, so I learn more about that -- and more abstract things like new programming languages.
Because it's harder to administrate windows filesharing as the usernames aren't synced between the machines, and there's a limit on how many people can be connected at once (10 for pro, 5 for home).
Samba can do more than that, and usernames are irrelevant -- remember box07, box08, etc? They all have access to everyone's files.
Does your plumber teach you how to plumb your house? Does your auto mechanic teach you how to fix your car?
I suspect my plumber would tell me some basic stuff -- like "I unclogged your pipes, but if you want them to stay that way, pour some Drano in every now and then." And, if I asked, they might tell me the best kind of Drano.
You sound like you're pretty young and expect someone to hold your hand through learning anything new.
It's actually not me that's the problem. I could learn all of this on my own, I'm just busy with other things, and not incredibly interested. It's the people in the office who need hand-holding -- and would very much like to be independent.
If you ever get a MacBook, you will literally be able to drag and drop most software from your PowerBook and expect it to run.
The PPC to x86 transition is the second one Apple has made, and the reason it works is that each time, they switched to such ludicrously faster processors that emulation (JIT compilation) still made programs run faster on the new systems. From what I hear, PPC programs run twice as fast on a MacBook Pro as a Powerbook, and Universal binaries run four times as fast.
So, that is how x86 would get replaced. Actually, Transmeta tried that -- they created their own architecture, and emulated x86 in firmware. Unfortunately, their chips are obscenely slow, and they haven't given us any compilers for their own arch, so it didn't work well -- but that's how you'd do it. Make your chip fast enough and your JIT good enough that it's actually not a performance loss to run x86 on <insert new architecture>, and people will switch. Which is sort of what's happening with x86_64.
But don't make Transmeta's mistake -- actually give us a compiler for it.
Only if they licensed it under the "GPLv2 or later", which most are. And not the other way around.
You can make changes, but you only own the copyright to your changes. Thus, you have to get permission from all the copyright holders in order to change the license. However, if it's licensed under "GPLv2 or later", then you can certainly distribute it under "GPLv2 only", or "GPLv3 only", or "GPLv3 or later".
By the way, this is why several large open source projects create organizations to hold the copyrights -- so they can change the license. If tons of people own the copyrights, but it's licensed under "GPLv2 or later", then you can only change the license to later versions of the GPL -- and anyone else can, too. This is also worrying, because the FSF can release any GPL they want, and your software can be released under it. However, if you keep the copyrights and release it under "GPLv2 only", then you can choose to switch to v3 later, or BSD, or any other license you like -- you can dual-license it.
So, basically, what MySQL is doing is saying that the MySQL team gets to choose the license, and the FSF does not. And there's nothing wrong with that. I do hope they go GPLv3, but if it was my software, I'd be waiting for the final version, too.
bad idea. this means when BluRay comes out, someone cannot take existing DVD-based GPL code and mod it to work with BluRay if it won't also work with DVD.
No.
As I understand it, this does not require any modified version to be able to play a DVD, any more than GPLv2 requires everyone who downloads VLC to download the VLC source code. It simply requires it to be possible.
Now, this does mean we can't really have GPL'd Blu-Ray players until it's broken. That is because if anyone ever wrote a GPLv3 Blu-Ray player, it would have to be possible for me to modify it to rip perfect copies of that disc. Thus, they might use GPLv2 with Blu-Ray, because they can use Trusted Computing to prevent any modified version from working, but they cannot use GPLv3.
The hope of GPLv3 is that it will be a factor in convincing the **AA's to stop this DRM crap.
(a) "if you bring suit" is not the same as "if you win the suit".
Correct. The FSF doesn't want patent suits to ever be an issue in software development, and frankly, I agree. It will likely be at least fifty years until any government implements a patent system that actually reflects the realities of software.
In the meantime, this essentially says that if you want to use the software, you cannot sue any of the developers for patent infringement. And that seems fair enough, even ignoring the fact that software patents don't work now. As it is, the only people I can imagine this causing problems for are patent trolls.
After all, if you have patented an idea, and you're actually developing useful software with it, there's nothing to stop you from using your own software instead of the GPL'd stuff. You're just not allowed to type your court papers in OpenOffice while you're suing OpenOffice.org.
(b) why would a patent suit end said permission?
Because the license says so. Now, if your question is "why should it"... see above and below.
if the patent suit is in one country, development could then just move to another country under which the original patent has no jurisdiction.
Except that there are treaties, and that there's a limited number of countries to go to, and that I'd have to give up development. C'mon, man, I'm eating Ramen, you expect me to move to Finland or something, just to avoid the patent trolls?
Fuck no. We make a stand, here, now. No more patent trolls -- or at the very least, they don't get to use my software.
A linux box maintained by a 15 yr old and his older brother usually leads to a poorly documented mess which neither you nor your brother understands after several years.
To some extent, yes. In my case, I recently came back to a Debian firewall that had bad RAM or something, and was so ridiculously out of date (with updates) and a bit corrupt from the RAM that I decided to rebuild it. And it had been years since I set it up.
However, mostly from memory, I was able to copy over config files and get a new Ubuntu firewall in one Saturday afternoon -- most of it was spent finding out which hardware was good, and eventually stealing the boss's desktop to make it work.
But, thank you for reminding me -- I will teach him to document everything.
Except for the OS upgrades, and the security upgrades, etc. pp. The same goes for windows servers, btw.;)
True enough, but I've been amazed at how little I have to do of that. The box under my desk has been running for months. Every now and then, I ssh in, apt-get update && apt-get dist-upgrade. Sometimes there's one or two updates -- and not a single update has broken anything, including the Dapper->Edgy upgrade.
So, yes, there's some of that, but it's almost a non-issue, especially with something simple like a Samba/OpenVPN box. Compare that to a Windows server, where you also have to keep IE up to date, and run it through a GUI, and reboot for just about every update...
However, in my experience, Windows servers do occasionally have actual problems to solve, whereas Linux servers just keep going until hardware failure -- such as the bad RAM I mentioned earlier.
Not using tapes has it's own set off problems (slow links for off-site storage).
True enough, but it doesn't take much. Keep in mind, too, that worst case, we can forgo the disk images (or put those on tapes) -- PowerPoint files over a slow link isn't a big deal, and if we lose the disk images, we rebuild, not a big deal compared to actual lost data.
It's not an usual approach to this problem, but i think it will work well, as long as the user don't have any important data stored locally.
That's the plan. And, knowing that the local disk will be wiped routinely will do wonders for users keeping their own discipline in this matter;)
Since they aren't currently backing up the local disk anyway, this means that if they lose anything from this procedure, they would've lost it sooner or later anyway. Obviously, we'll try to prevent that happening. And the positive is, if they don't lose a thing, they have the peace of mind of knowing the backup is working flawlessly.
A full, complete, solid customer installation for 10 workplaces usually comes up to 58000US$.
By "workstation", I mean "desktop computer". I'm talking about 10 people, not 10 offices. And if they do scale up dramatically enough for the domain to make a difference, they'll have bigger problems anyway.
Linux has it's place - i use it for webhosting (a side job) and of course communications infrastructure.
But what is it that makes it inappropriate here? Consider:
Central managment of updates, virus scanners, software, images, user data, backup
Central management of updates can be managed with a local repository or any number of custom scripts, or you could use some diskless monstrosity -- with the new fscache stuff, this is starting to look pretty attractive, just install the updates on the master and watch all the servers automatically be updated.
Virus scanners aren't necessary unless you've got Windows boxes, and if you do, you can filter them at the firewall, even a proxy server. ClamAV is pretty decent, and manages its own updates.
No, it does not require that all modified versions be able to do that. It simply requires that a modified version can be written to do that.
This means, for instance, Microsoft can't take Gaim, release it as a new version of MSN Messenger, and use Trusted Computing to prevent their servers from communicating with modified versions of Gaim.
We are unable to contact some contributors to get their ok on using GPLv3, and rather than disrespect their contributions by pushing the bottom line of v3, we're going to have to keep using v2 since it's the license they submitted with.
I agree with others here -- even if it's legally OK for you to do that, if they released their code under v2 or later, then they have to be OK with it. You'll have to decide for yourself whether you want v3 or not -- and I'd suggest you either wait for v3 to be released, or make yourself a part of the v3 process. They have a public site where they're holding all their discussions.
v3 is about pushing an agenda within a license from what I can tell, rather than sticking to what it is, a license.
And v2 wasn't an "agenda"? Every license has an agenda, even no license (public domain) is its own agenda. You may not agree with what the FSF is trying to do here, but if so, have the balls to say that, and why you disagree. Don't hide behind "It's an agenda" and "We can't because we want to respect these developers we've lost touch with."
By the way, if you read the FSF's philosophy pages, v2's agenda is identical to v3's, and indeed, identical to v1's. That agenda is, anywhere someone can get a copy of your software running, they should be able to see and modify the source code, and run that. The only difference is, v2 had some loopholes that v3 doesn't. You may like these loopholes, but that doesn't change what they are -- bugs in the license.
If that isn't what you intended, I'm curious as to why you didn't go with a BSD license.
Consider the Linux kernel. They now cannot use any license other than GPLv2.
Now, consider the Gentoo project. As far as I know, every single ebuild is copyrighted the "Gentoo Foundation". That means they can relicense it to anything they want, at any time.
If I were to keep all the rights to all my code -- and any contributions to it -- I'd certainly be happy with saying GPLv2 only, until I'm convinced I like GPLv3. Otherwise, if it's something like Linux, where every contributor owns the rights to their own code, "or later" makes sense in that it gives someone the authority to change it later, and the FSF is certainly not the worst organization you could trust with that authority.
All things considered, I'd rather own my own stuff, but realistically, you need to be able to change the license somehow -- the GPLv2 will eventually have to be replaced. People will find loopholes -- hell, they already have, with TiVo, but maybe the next loophole will be one you care about. So, unless you can have an individual or an organization own all the rights to a given piece of software, your only way to ever change the license is that "or later" clause.
Except that here, no one knows how to do it, so it costs them a pile of money to setup a new machine.
That's the Tech's fault - why?
When someone buys something they didn't really want, they only have themselves to blame.
It just strikes me as unethical. These techs know as well as I do what would really work, and what wouldn't -- or they're more clueless than I thought.
Let me put it this way: If someone comes in to your computer shop wanting to buy a new video card to make their spreadsheet run faster, do you at least point out that video cards generally don't do that? At least try to educate them -- and you might be wrong, too (video cards can be used for some math), but at least try not to rip them off, even if it would be their fault.
Ouch. And they still hire those guys? I mean, Windows 2000 is no longer supportet by MS - e.g. Timezone changes.
To be fair, this was someone in-house who wanted to stay with something "familiar". The only thing that convinced them was when a second laptop came in and there was no time to wipe it -- and guess which laptop works better now?
with rsync or DRBD backups over the Internet to someone's house
Don't do Online-Backups unless you always have technical people on site. Tapes are something users can grasp much better - they are much better at dealing with them, and you get less problems.
Actually, this is half the reason for doing online backups -- no manual interaction. It just goes, and they can browse the backup with a web browser if they want to make sure it's working.
Which is irrelevant when these cunts won't teach us how to admin it.
Probably because that's not their job. I'm not a good teacher either, that's why i'm doing a tech job and not a teaching job. If you lack windows knowledge, go learn it. All they need to do is document their setup.
"Documenting their setup"... I wonder if they do that, I doubt it. But seriously, how hard is it to write step-by-step instructions for setting up box11 when box10 goes down? Or point them to some documentation on how to do that?
Don't get me wrong, but you sound like a guy who wouldn't have much problems with learning all the windows stuff - but you don't want to.
That much is true -- mostly because I think it's unnecessary. If I could convince them to treat it as a domain, I'd admin it as a domain, but I really doubt I can do that -- there's not really much discipline among users, there.
I'm not trying to say that your brother is incompetent, but one thing is for sure: he lacks experience.
Let me put it this way: First, he's 15 years old, so $10/hour is significant, especially when I'm still doing much of the work. Second, this is a way for him to get that experience. And finally, he'll mostly be doing the kind of stuff I could train anyone in that office to do -- for instance, take an image of a hard drive, swap it out for a new drive, restore the image from the fileserver. I imagine that it will eventually become his job to format and re-image the machines once a month.
If there are problems, I'm not that far away -- I just don't have time to babysit their infrastructure, I have a real job and another freelance gig, so I am teaching him to take over the more mindless stuff. Which is, by the way, how I got started -- an older, more experienced admin gave me the root password to a Linux shop, gave me the boring stuff like tweaking and compiling custom kernels, and taught me what I know about admining Debian systems.
Regardless, it's not just me. Everyone and their dog knows how to right-click and go "Sharing".
Yeah, but users shouldnt do that. That's better be left to IT staff.
It's called "Active Directory" or "Windows Domain".
And it's also called an NT domain -- or a Windows NT Domain, if you prefer.
Setting up a new machine is easier,
Except that here, no one knows how to do it, so it costs them a pile of money to setup a new machine.
users get their old profile synced back from the server,
Sounds great, except it doesn't happen here. Each machine has a name like "box7" or "box10", with a username/password based on that. No one will ever login to "box7" except from on a machine called box7.
I realize this isn't Active Directory's fault, I'm just pointing out that as far as I can tell, Active Directory was only used here so these cunts can charge thousands of dollars to come over and twiddle their thumbs while they install Win2K on brand new Dell laptops.
all "my documents" are properly redirected to the server,
Which can be setup manually, easily. Perhaps not as easily as Active Directory does it, but we all know how to right-click My Documents.
and users will have less hassle.
Right now, the user experience is identical to before Active Directory, except that they now have to type a password on boot, and nobody in-house knows how to maintain the thing. They went from one overkill solution that they didn't know how to maintain (Novell) to another (Active Directory).
Setting up a new machine can be automated completely using RIS (XP) or WDS (Vista)
I can automate that with nLite or with Linux disk images, but more to the point, it is currently NOT automated. Most of the machines there are Win2K boxes, and whenever one gets so full of spyware that no one knows what to do with it anymore, it has to be wiped and reinstalled, and they have to shell out another few hundred dollars to The Techs.
This is actually the only thing that keeps us going back to them. Everyone in the office would much rather use me or my brother -- we're cheap and efficient, and we don't try to sell them useless tech. So, eventually, I'm thinking we'll replace it all with a Samba server running FileMaker Server either natively or on Wine, with rsync or DRBD backups over the Internet to someone's house. The only other need for a Windows server is RDP, but we could get rid of that with a real and properly configured VPN -- which would be child's play if the fileserver was a Samba running on the VPN box.
That's a selling point, but usually the main point is less administrative hassle.
Which is irrelevant when these cunts won't teach us how to admin it. Much the same as I'd call Ubuntu "less administrative hassle," but that's irrelevant when no one there has touched Linux, or wants to. Still, I think tying them to a 15-year-old brother for $10/hour is better than two MSCE bastards for $75/hour.
You blame your incompetency on others?
Lack of knowledge and interest.
Regardless, it's not just me. Everyone and their dog knows how to right-click and go "Sharing". And if it's as easy as you say it is, I'm now tempted to learn it so I, too, can be paid $75/hour for doing nothing.
and it runs over a VPN because RDP doesn't do crypto.
You're wrong.
Oh. My bad. It runs over a VPN for no reason whatsoever, other than that I first sold them on the idea of a VPN.
Yes, I know. And Lord of the Rings was intended to be a single book, so it got the sequels because that's how big the story was, not because of popular demand after the first one.
And yet, my mother only saw the second one -- she missed the first, and after the long, brutal, tiresome battles of the second, she didn't want to see the third.
If everyone had taken that attitude, then there wouldn't have been a third, even if that's how the books go. And we all know they differ from the books anyway (where's Tom Bombadil?)
Since some people seem to be sharing horror stories, I'll share mine. Kind of tame...
This small company doesn't have an IT department, they have a person who finds time in her day to do IT -- maybe. I occasionally get to come in and try to clean up the mess.
This is a small business. By "Small", I mean "Roughly ten, and no more than twenty people in the company." By that, I mean that there cannot be more than about ten workstations, all running Windows 2000 or above. At least they aren't still on Win98 and Win95...
They have a server -- a fileserver, on which all their data is stored, which also runs the FileMaker server, and is (theoretically) backed up nightly. So far, so good, even if it is done with Microsoft Backup.
So, not good, but not horrible.
Here's the problem: They hire some local IT company, at a rate of some $75/hour/tech for two techs, to handle just about anything that goes wrong, except on rare occasions where I convince them I can do it quicker and cheaper. These guys have set them up with a fucking NT domain!
Would someone please explain to me why the fuck a company that small needs an NT domain?
I mean, they aren't swapping computers -- EVER. Theoretically, everything's backed up, but they never test that -- everyone has their POP3 email downloaded locally to Outlook, which is going to bite them in the ass someday. Everyone has access to everyone else's files on the server, to make sharing stuff easier. The office is small enough, and has no wireless to speak of, so it's not as if anyone's going to be snooping around their network.
I just don't see why their needs couldn't be met with simple Windows filesharing, or a Samba server.
Now, why not have an NT domain, especially if they've already paid for the "Server Edition" or whatever? Well, it costs somewhere between $200 and $1000 of tech time to set up a new computer to operate with "the network" -- which basically means, install printer drivers, get it onto the domain, map a couple of drives, and move "My Documents" to the fileserver. I do not know how to admin an NT domain, so I cannot do this myself.
This also means, no reinstalls, really, because a reinstall ends up costing almost as much as a new computer.
And, of course, the techs refuse to teach any of us how to admin our own fucking network, because if they did that, they'd be out of a job. That is why it's a problem. My 15-year-old brother can work dirt-cheap and setup Windows filesharing, and even Samba servers, all day long, but neither of us knows what to do with an NT domain -- but these techs have convinced them that it's somehow "more secure".
That's the one thing that jumps out of me. I could make a laundry list of other offenders -- just about every machine there is dog-slow (likely spyware), almost everytime I sit down at someone's computer there's some 30 critical updates waiting, they've been known to do things like wipe XP Pro off a laptop and put 2K Pro on in order to have something "more familiar" (they've since learned from that mistake)...
I remember offering to setup a VPN for them, a simple affair with a Linux box and OpenVPN, figuring I'd just route things like Samba and FileMaker connections. And it worked, flawlessly, except I couldn't get on "the network", because I couldn't make the NT domain go across the VPN. And so, within a day or so of my attepmt, they called in The Techs to set them up a VPN (since they'd heard from me that VPNs are a good thing), and now they use Remote Desktop and call it a VPN. True, they can do what I was promising -- they can access their email, files, and FileMaker database remotely -- but it's also damned inefficient, and it runs over a VPN because RDP doesn't do crypto.
Anyway, bit of a rant, but if there's a moral, it's this: If you can't afford a full IT person, you certainly can't afford freelance MSCE dicks.
Slashdot Mirror
You make the point well enough about all the security. That means it's a lot saner to run your web configuration through Apache or something. It also means you'll have to deal with issues like a username/password, and you'll probably want to proxy the whole thing through SSH or something (or you'll need an SSL-enabled Apache)...
Unless you're doing something AJAX-y, changes do require you to fill out a form and hit "submit", and wait for a response. I don't really see how this is different than editing a config file and pinging the app. But if you're going to do something AJAX-y, we're getting pretty heavy into the realm of external dependencies -- this is no longer something you want embedded into your app, the way CUPS configuration is.
And I don't really want to have to install Apache, mod_ssl, and some AJAX libs just to configure, say, an IRC server, or a DNS server.
Besides: It doesn't have to require pinging the app. I seem to remember exim checking the timestamp on its config file with every mail received, so if you made a change, you could expect it to take effect instantly. Another way to do this would be inotify -- I believe vim writes all your changes to a temporary file, then 'rename's the temporary file on top of the original one. Thus, as soon as you see a new file there, you can probably parse it -- worst case, if you don't trust the text editor, you can wait a second or two for the inotify events to settle.
Now, there are advantages -- you can actually display a finite set of options, you can show the user what their system can actually handle, and there aren't going to be any syntax errors. But none of these have ever really been a problem for me -- I can crack open just about any config file, read the comments, and make a few changes, and be confident that it'll work.
I think this is pretty much how it's going to be from now on.
Major innovation will be on Linux and OS X. It will take a month or two for Linux to absorb new OS X features, and OS X will generally absorb some of the good ideas from Linux in the next release of their OS, which seems like it's going to be every couple of years. Linux will always have more features than anything else, and OS X will always be perceived as "easier" in some way.
Windows will always be just slightly behind the last version of OS X. That is, Vista and Leapord come out, and Vista will be almost as good as Tiger, with Leapord completely blowing it away. And I'll take the features I like from both, and implement them in Linux, and still have the features neither will touch, like a real package manager.
In any case, I think it's more compelling to make that kind of comparison. For instance, if we were able to show that Ubuntu Dapper is better than XP, that'd be a very compelling argument, because we could then say "Oh, by the way, we're on Edgy now, but XP can't even begin to compete with that."
I'll have to check that, but it's doubtful. This is wireless, so Gigabit just doesn't make sense. I do occasionally use a Gigabit crossover...
But so what? Does this mean we can "ping of death" these crap Linksys boxes? Gives new meaning to "wardriving"...
Yes, it was the MM10. I didn't trust Sharp again, though, for a couple of major reasons:
Now, I've discovered how much I dislike certain aspects of the Powerbook, and I could be talked into getting another Actius, or something like it. But this time around, I want it to be able to play DVD video -- the MM10 choked on that, either the CPU or the video card, but it was not just Linux -- and I want it to have at least 40 or 50 gigs of storage, unless it's all solid state. Other than that, it's software (Linux) issues that I can fix myself...
That's making a big assumption -- that this COBOL code all works, and that no one needs to change it much. That may be true for a lot of things, but as soon as you actually have to maintain the fucker, it starts actually looking cheaper to rewrite it -- even in Java -- than to try to tack on the features you need to the COBOL monstrosity.
My Powerbook -- well, the screen is dead for the moment, but it has been a router killer for some time now. I haven't been able to figure out whether it's OS X, the VPN, the SSH, or what, but everywhere I go with this thing, routers die and have to be reset (pull the plug). Sometimes it doesn't happen for days, sometimes it happens every hour or so, sometimes I open the thing up from sleep, get all connected to the wireless, and watch it kill the router.
I'm hoping that Vista will convince these router manufacturers to get their act together. Even if it is somehow OS X's fault, or my fault, I shouldn't be able to DOS a router that easily.
Except that it's easy to set up (there's even a GUI for that now, I think), and it's about what they're used to. If I was going to really change things around there, it would be neither Samba nor domain controllers, it'd be a Linux fileserver, a ton of Linux desktops, and a single Windows application server. But that would be a huge project, and I can't make a good business case for it -- and there isn't too much short of that to hold my interest.
True enough. In this case, though, all they need to do is be able to setup a new box (and account), and probably tear down the old account and permissions. According to you, that's actually pretty easy -- so we give them a printed list of step-by-step instructions.
Ideally, of course, I'd have a script to handle this for them. Input the name and mac/ip address of the new box, then a username and password, then: done. But it really does sound like that's about what you do anyway.
Second, you're right, old format -> OOXML -> old format... yeah, that would kind of suck. However, I don't get why you would ever want such a loop to work. Old format -> OOXML -> some other format -- now that makes sense. Think of it this way -- you can burn your mp3s as music CDs, then rip those CDs into mp3s, but you're going to lose something.
But keep in mind: If OOXML -> old format is to work, it must be able to cover anything OOXML might do, including hacks to make the old format look right.
And more to the point, have you actually looked at these? Some of them seem entirely too simple: Emulate WordPerfect 5.x line spacing. Really, how hard would it be to simply have a generic <p spacing="0.8"> (just an example, I don't know how this is done in practice or exactly what the difference is) in your conversion of WordPerfect to OOXML? And, going the other way, couldn't you just perform a similar calculation to determine line spacing?
I mean, you're right, it takes a bit of effort. Yeah, your conversion tool takes a little more effort to write than it otherwise would. I think that's an acceptable loss, considering it saves a massive amount of effort from every fucking word processor to implement this standard, EVER.
As for some of this other stuff, some of it seems inevitable, if you want the document to look perfect. And if you accept that, then yes, you need these tags -- but would it've killed Microsoft, in their 6000 word "spec", to spell out what an app's actually supposed to do? Specs are supposed to be the basis for implementations -- they should not have to refer you to another implementation -- and it certainly doesn't help Microsoft's case of "open"-ness when it would cost me a significant amount of money to obtain all the word processors they refer to.
I, for one, don't accept that. The vast majority of these tags seem to be referring to specific weird behavior, like character spacing being slightly off, when they could've simply given us an option to specify how much space between each character. That is, whether it's for "backwards compatibility" or not, you should be able to specify... oh, let's say: Rather than... say: Now, it's true, the second one takes more space. That's why we use compression techniques -- both actual zip compression and programmatic compression. For instance, we could do something like this: Worst case, you can use XML entities.
Now, you're right, half-baked conversion tools will screw you over -- but so will half-baked software. And really, after any conversion like this, doesn't it make sense to go back and check that the document was converted properly before deleting the original file?
The point of a new standard is not to support tons of legacy documents and software. It's to make sure we're all on the same page moving forward, and Microsoft's 6000 page monstrosity doesn't help much with that, especially when it's not even complete till you buy and reverse engineer 16 years worth of word processors.
Please give us direct access to the device -- none of this CompactFlash HD emulating crap. Windows can emulate a hard drive if it wants to -- Linux is developing an FS specifically designed for Flash devices, that operates directly on the Flash, not even as a block device.
Remember -- Flash is funny. You don't overwrite, you erase, then write -- in fairly large blocks. You want to intentionally fragment stuff, so you write evenly over the memory, so you don't wear it out as quickly -- but you also want to keep the chunks defragmented, even delay writes a little (Reiser4/XFS style), so you can overwrite a whole block at a time.
All in all, I'd much rather trust Linux/jffs2 to manage my Flash than some arbitrary hardware inside a CompactFlash device.
And speaking of OS choices... Why are you so desperate for OSX on such a device? I mean, you realize you've just reduced yourself to grovelling because your OS vendor needlessly restricts the hardware you run it on? You realize that the moment someone produces such a device, I'll be able to run Linux on it, but unless you're very lucky, you won't be able to run OS X without cracking it?
Regarding battery life -- my old Sharp Actius had some 9-10 of battery life, and it was a subnotebook. If you want battery life, it's simple: Sacrifice performance, and don't do much with it (read a book or something). Unfortunately, it doesn't seem like any manufacturer wants to do that.
If I remember right, PS3 Linux (or any other OS) runs in a hypervisor. That means virtualization -- it means that they actually can prevent Linux from gaining access to the hardware, just as I can run Windows safely inside qemu on my Linux and deny it access to my hardware.
So, it's not just that there are no drivers, and that you can hope someday someone will reverse-engineer it enough -- no. They'd have to crack it and reverse-engineer it -- basically like any kind of Xbox Linux.
As far as whether HD video works, I haven't gotten a straight answer from anyone. We know 3D acceleration doesn't work, and we know you get a framebuffer. What I don't know is whether the framebuffer is accelerated enough to play HD video or 2D games. But keep in mind -- even if it can play HD video, it seems doubtful that you'd actually be able to play a Blu-Ray disc.
No, actually, they wrote new software that runs on Linux, as I recall. Nothing to stop you from releasing proprietary software which runs on Linux. But, they do prevent us from actually using the code they did modify -- and by the way, they do give us the code, they just don't let us use it.
Nothing in GPL, either 2 or 3, prevents you from protecting it from plagarism. But I don't think "plagarism" means what you think it means.
So what? What does this have to do with anything?
First off -- TiVo is a piece of hardware. Second, others have indeed done TiVo clones, some with Linux, some without.
But more relevantly, there is nothing in GPLv3 which prevents you from making a professional, user friendly product. You certainly can -- and it will be all the more user friendly for me if I can actually modify your code.
Besides, if you truly believe open source would be a hobbyist affair without tivoization, you should look up Ubuntu, RedHat Enterprise Linux, and the vast numbers of Debian servers around the world. Linux doesn't need TiVo, and it doesn't need you. After all, what's the point in letting TiVo use it, if we get nothing back?
One more thing to remember: You're getting the code for free. You do not get to bitch about us changing the terms. Nothing is stopping you from developing your own TiVo DRM monstrosity, but you will not be doing it with my code -- at least, not for free.
Hmm. Looks about right. If I read that link right, it's comparing PPC-on-Rosetta-on-x86 to native x86.
My point was, Rosetta runs slower than native, of course. But, the x86 processor is itself much faster than a PPC processor. Not because of the architecture -- because of raw speed. Thus, even if you're running entirely PPC apps, a MacBook can still feel like an upgrade -- and of course, everyone recompiles for x86 (or "universal") quickly enough, because the x86 ones are that much faster.
What we'd want to avoid is feeling an actual loss in speed when switching to the new arch. In order to beat Worse Is Better, you have to be much better, enough that it's really a compelling switch. That's why x86_64 works -- it is a lot better (additional registers, more RAM), not much more expensive, and has a perforamnce hit of exactly zero for 32-bit code.
Yes, you're absolutely right -- and if you read my post carefully, you'll find that's roughly what I said. Except for the part about it being 6 books... that's a technicality, anyway.
On the contrary, I learn new things all the time -- just things I like to think of as useful. I don't want to learn Active Directory, because I'd much rather set people up with Linux, so I learn more about that -- and more abstract things like new programming languages.
Samba can do more than that, and usernames are irrelevant -- remember box07, box08, etc? They all have access to everyone's files.
I suspect my plumber would tell me some basic stuff -- like "I unclogged your pipes, but if you want them to stay that way, pour some Drano in every now and then." And, if I asked, they might tell me the best kind of Drano.
It's actually not me that's the problem. I could learn all of this on my own, I'm just busy with other things, and not incredibly interested. It's the people in the office who need hand-holding -- and would very much like to be independent.
If you ever get a MacBook, you will literally be able to drag and drop most software from your PowerBook and expect it to run.
The PPC to x86 transition is the second one Apple has made, and the reason it works is that each time, they switched to such ludicrously faster processors that emulation (JIT compilation) still made programs run faster on the new systems. From what I hear, PPC programs run twice as fast on a MacBook Pro as a Powerbook, and Universal binaries run four times as fast.
So, that is how x86 would get replaced. Actually, Transmeta tried that -- they created their own architecture, and emulated x86 in firmware. Unfortunately, their chips are obscenely slow, and they haven't given us any compilers for their own arch, so it didn't work well -- but that's how you'd do it. Make your chip fast enough and your JIT good enough that it's actually not a performance loss to run x86 on <insert new architecture>, and people will switch. Which is sort of what's happening with x86_64.
But don't make Transmeta's mistake -- actually give us a compiler for it.
Only if they licensed it under the "GPLv2 or later", which most are. And not the other way around.
You can make changes, but you only own the copyright to your changes. Thus, you have to get permission from all the copyright holders in order to change the license. However, if it's licensed under "GPLv2 or later", then you can certainly distribute it under "GPLv2 only", or "GPLv3 only", or "GPLv3 or later".
By the way, this is why several large open source projects create organizations to hold the copyrights -- so they can change the license. If tons of people own the copyrights, but it's licensed under "GPLv2 or later", then you can only change the license to later versions of the GPL -- and anyone else can, too. This is also worrying, because the FSF can release any GPL they want, and your software can be released under it. However, if you keep the copyrights and release it under "GPLv2 only", then you can choose to switch to v3 later, or BSD, or any other license you like -- you can dual-license it.
So, basically, what MySQL is doing is saying that the MySQL team gets to choose the license, and the FSF does not. And there's nothing wrong with that. I do hope they go GPLv3, but if it was my software, I'd be waiting for the final version, too.
No.
As I understand it, this does not require any modified version to be able to play a DVD, any more than GPLv2 requires everyone who downloads VLC to download the VLC source code. It simply requires it to be possible.
Now, this does mean we can't really have GPL'd Blu-Ray players until it's broken. That is because if anyone ever wrote a GPLv3 Blu-Ray player, it would have to be possible for me to modify it to rip perfect copies of that disc. Thus, they might use GPLv2 with Blu-Ray, because they can use Trusted Computing to prevent any modified version from working, but they cannot use GPLv3.
The hope of GPLv3 is that it will be a factor in convincing the **AA's to stop this DRM crap.
Correct. The FSF doesn't want patent suits to ever be an issue in software development, and frankly, I agree. It will likely be at least fifty years until any government implements a patent system that actually reflects the realities of software.
In the meantime, this essentially says that if you want to use the software, you cannot sue any of the developers for patent infringement. And that seems fair enough, even ignoring the fact that software patents don't work now. As it is, the only people I can imagine this causing problems for are patent trolls.
After all, if you have patented an idea, and you're actually developing useful software with it, there's nothing to stop you from using your own software instead of the GPL'd stuff. You're just not allowed to type your court papers in OpenOffice while you're suing OpenOffice.org.
Because the license says so. Now, if your question is "why should it"... see above and below.
Except that there are treaties, and that there's a limited number of countries to go to, and that I'd have to give up development. C'mon, man, I'm eating Ramen, you expect me to move to Finland or something, just to avoid the patent trolls?
Fuck no. We make a stand, here, now. No more patent trolls -- or at the very least, they don't get to use my software.
To some extent, yes. In my case, I recently came back to a Debian firewall that had bad RAM or something, and was so ridiculously out of date (with updates) and a bit corrupt from the RAM that I decided to rebuild it. And it had been years since I set it up.
However, mostly from memory, I was able to copy over config files and get a new Ubuntu firewall in one Saturday afternoon -- most of it was spent finding out which hardware was good, and eventually stealing the boss's desktop to make it work.
But, thank you for reminding me -- I will teach him to document everything.
True enough, but I've been amazed at how little I have to do of that. The box under my desk has been running for months. Every now and then, I ssh in, apt-get update && apt-get dist-upgrade. Sometimes there's one or two updates -- and not a single update has broken anything, including the Dapper->Edgy upgrade.
So, yes, there's some of that, but it's almost a non-issue, especially with something simple like a Samba/OpenVPN box. Compare that to a Windows server, where you also have to keep IE up to date, and run it through a GUI, and reboot for just about every update...
However, in my experience, Windows servers do occasionally have actual problems to solve, whereas Linux servers just keep going until hardware failure -- such as the bad RAM I mentioned earlier.
True enough, but it doesn't take much. Keep in mind, too, that worst case, we can forgo the disk images (or put those on tapes) -- PowerPoint files over a slow link isn't a big deal, and if we lose the disk images, we rebuild, not a big deal compared to actual lost data.
That's the plan. And, knowing that the local disk will be wiped routinely will do wonders for users keeping their own discipline in this matter ;)
Since they aren't currently backing up the local disk anyway, this means that if they lose anything from this procedure, they would've lost it sooner or later anyway. Obviously, we'll try to prevent that happening. And the positive is, if they don't lose a thing, they have the peace of mind of knowing the backup is working flawlessly.
By "workstation", I mean "desktop computer". I'm talking about 10 people, not 10 offices. And if they do scale up dramatically enough for the domain to make a difference, they'll have bigger problems anyway.
But what is it that makes it inappropriate here? Consider:
Central management of updates can be managed with a local repository or any number of custom scripts, or you could use some diskless monstrosity -- with the new fscache stuff, this is starting to look pretty attractive, just install the updates on the master and watch all the servers automatically be updated.
Virus scanners aren't necessary unless you've got Windows boxes, and if you do, you can filter them at the firewall, even a proxy server. ClamAV is pretty decent, and manages its own updates.
User data and backup is the same as on Wi
No, it does not require that all modified versions be able to do that. It simply requires that a modified version can be written to do that.
This means, for instance, Microsoft can't take Gaim, release it as a new version of MSN Messenger, and use Trusted Computing to prevent their servers from communicating with modified versions of Gaim.
I agree with others here -- even if it's legally OK for you to do that, if they released their code under v2 or later, then they have to be OK with it. You'll have to decide for yourself whether you want v3 or not -- and I'd suggest you either wait for v3 to be released, or make yourself a part of the v3 process. They have a public site where they're holding all their discussions.
And v2 wasn't an "agenda"? Every license has an agenda, even no license (public domain) is its own agenda. You may not agree with what the FSF is trying to do here, but if so, have the balls to say that, and why you disagree. Don't hide behind "It's an agenda" and "We can't because we want to respect these developers we've lost touch with."
By the way, if you read the FSF's philosophy pages, v2's agenda is identical to v3's, and indeed, identical to v1's. That agenda is, anywhere someone can get a copy of your software running, they should be able to see and modify the source code, and run that. The only difference is, v2 had some loopholes that v3 doesn't. You may like these loopholes, but that doesn't change what they are -- bugs in the license.
If that isn't what you intended, I'm curious as to why you didn't go with a BSD license.
Consider the Linux kernel. They now cannot use any license other than GPLv2.
Now, consider the Gentoo project. As far as I know, every single ebuild is copyrighted the "Gentoo Foundation". That means they can relicense it to anything they want, at any time.
If I were to keep all the rights to all my code -- and any contributions to it -- I'd certainly be happy with saying GPLv2 only, until I'm convinced I like GPLv3. Otherwise, if it's something like Linux, where every contributor owns the rights to their own code, "or later" makes sense in that it gives someone the authority to change it later, and the FSF is certainly not the worst organization you could trust with that authority.
All things considered, I'd rather own my own stuff, but realistically, you need to be able to change the license somehow -- the GPLv2 will eventually have to be replaced. People will find loopholes -- hell, they already have, with TiVo, but maybe the next loophole will be one you care about. So, unless you can have an individual or an organization own all the rights to a given piece of software, your only way to ever change the license is that "or later" clause.
It just strikes me as unethical. These techs know as well as I do what would really work, and what wouldn't -- or they're more clueless than I thought.
Let me put it this way: If someone comes in to your computer shop wanting to buy a new video card to make their spreadsheet run faster, do you at least point out that video cards generally don't do that? At least try to educate them -- and you might be wrong, too (video cards can be used for some math), but at least try not to rip them off, even if it would be their fault.
To be fair, this was someone in-house who wanted to stay with something "familiar". The only thing that convinced them was when a second laptop came in and there was no time to wipe it -- and guess which laptop works better now?
Actually, this is half the reason for doing online backups -- no manual interaction. It just goes, and they can browse the backup with a web browser if they want to make sure it's working.
"Documenting their setup"... I wonder if they do that, I doubt it. But seriously, how hard is it to write step-by-step instructions for setting up box11 when box10 goes down? Or point them to some documentation on how to do that?
That much is true -- mostly because I think it's unnecessary. If I could convince them to treat it as a domain, I'd admin it as a domain, but I really doubt I can do that -- there's not really much discipline among users, there.
Let me put it this way: First, he's 15 years old, so $10/hour is significant, especially when I'm still doing much of the work. Second, this is a way for him to get that experience. And finally, he'll mostly be doing the kind of stuff I could train anyone in that office to do -- for instance, take an image of a hard drive, swap it out for a new drive, restore the image from the fileserver. I imagine that it will eventually become his job to format and re-image the machines once a month.
If there are problems, I'm not that far away -- I just don't have time to babysit their infrastructure, I have a real job and another freelance gig, so I am teaching him to take over the more mindless stuff. Which is, by the way, how I got started -- an older, more experienced admin gave me the root password to a Linux shop, gave me the boring stuff like tweaking and compiling custom kernels, and taught me what I know about admining Debian systems.
And it's also called an NT domain -- or a Windows NT Domain, if you prefer.
Except that here, no one knows how to do it, so it costs them a pile of money to setup a new machine.
Sounds great, except it doesn't happen here. Each machine has a name like "box7" or "box10", with a username/password based on that. No one will ever login to "box7" except from on a machine called box7.
I realize this isn't Active Directory's fault, I'm just pointing out that as far as I can tell, Active Directory was only used here so these cunts can charge thousands of dollars to come over and twiddle their thumbs while they install Win2K on brand new Dell laptops.
Which can be setup manually, easily. Perhaps not as easily as Active Directory does it, but we all know how to right-click My Documents.
Right now, the user experience is identical to before Active Directory, except that they now have to type a password on boot, and nobody in-house knows how to maintain the thing. They went from one overkill solution that they didn't know how to maintain (Novell) to another (Active Directory).
I can automate that with nLite or with Linux disk images, but more to the point, it is currently NOT automated. Most of the machines there are Win2K boxes, and whenever one gets so full of spyware that no one knows what to do with it anymore, it has to be wiped and reinstalled, and they have to shell out another few hundred dollars to The Techs.
This is actually the only thing that keeps us going back to them. Everyone in the office would much rather use me or my brother -- we're cheap and efficient, and we don't try to sell them useless tech. So, eventually, I'm thinking we'll replace it all with a Samba server running FileMaker Server either natively or on Wine, with rsync or DRBD backups over the Internet to someone's house. The only other need for a Windows server is RDP, but we could get rid of that with a real and properly configured VPN -- which would be child's play if the fileserver was a Samba running on the VPN box.
Which is irrelevant when these cunts won't teach us how to admin it. Much the same as I'd call Ubuntu "less administrative hassle," but that's irrelevant when no one there has touched Linux, or wants to. Still, I think tying them to a 15-year-old brother for $10/hour is better than two MSCE bastards for $75/hour.
Lack of knowledge and interest.
Regardless, it's not just me. Everyone and their dog knows how to right-click and go "Sharing". And if it's as easy as you say it is, I'm now tempted to learn it so I, too, can be paid $75/hour for doing nothing.
Oh. My bad. It runs over a VPN for no reason whatsoever, other than that I first sold them on the idea of a VPN.
Yes, I know. And Lord of the Rings was intended to be a single book, so it got the sequels because that's how big the story was, not because of popular demand after the first one.
And yet, my mother only saw the second one -- she missed the first, and after the long, brutal, tiresome battles of the second, she didn't want to see the third.
If everyone had taken that attitude, then there wouldn't have been a third, even if that's how the books go. And we all know they differ from the books anyway (where's Tom Bombadil?)
Since some people seem to be sharing horror stories, I'll share mine. Kind of tame...
This small company doesn't have an IT department, they have a person who finds time in her day to do IT -- maybe. I occasionally get to come in and try to clean up the mess.
This is a small business. By "Small", I mean "Roughly ten, and no more than twenty people in the company." By that, I mean that there cannot be more than about ten workstations, all running Windows 2000 or above. At least they aren't still on Win98 and Win95...
They have a server -- a fileserver, on which all their data is stored, which also runs the FileMaker server, and is (theoretically) backed up nightly. So far, so good, even if it is done with Microsoft Backup.
So, not good, but not horrible.
Here's the problem: They hire some local IT company, at a rate of some $75/hour/tech for two techs, to handle just about anything that goes wrong, except on rare occasions where I convince them I can do it quicker and cheaper. These guys have set them up with a fucking NT domain!
Would someone please explain to me why the fuck a company that small needs an NT domain?
I mean, they aren't swapping computers -- EVER. Theoretically, everything's backed up, but they never test that -- everyone has their POP3 email downloaded locally to Outlook, which is going to bite them in the ass someday. Everyone has access to everyone else's files on the server, to make sharing stuff easier. The office is small enough, and has no wireless to speak of, so it's not as if anyone's going to be snooping around their network.
I just don't see why their needs couldn't be met with simple Windows filesharing, or a Samba server.
Now, why not have an NT domain, especially if they've already paid for the "Server Edition" or whatever? Well, it costs somewhere between $200 and $1000 of tech time to set up a new computer to operate with "the network" -- which basically means, install printer drivers, get it onto the domain, map a couple of drives, and move "My Documents" to the fileserver. I do not know how to admin an NT domain, so I cannot do this myself.
This also means, no reinstalls, really, because a reinstall ends up costing almost as much as a new computer.
And, of course, the techs refuse to teach any of us how to admin our own fucking network, because if they did that, they'd be out of a job. That is why it's a problem. My 15-year-old brother can work dirt-cheap and setup Windows filesharing, and even Samba servers, all day long, but neither of us knows what to do with an NT domain -- but these techs have convinced them that it's somehow "more secure".
That's the one thing that jumps out of me. I could make a laundry list of other offenders -- just about every machine there is dog-slow (likely spyware), almost everytime I sit down at someone's computer there's some 30 critical updates waiting, they've been known to do things like wipe XP Pro off a laptop and put 2K Pro on in order to have something "more familiar" (they've since learned from that mistake)...
I remember offering to setup a VPN for them, a simple affair with a Linux box and OpenVPN, figuring I'd just route things like Samba and FileMaker connections. And it worked, flawlessly, except I couldn't get on "the network", because I couldn't make the NT domain go across the VPN. And so, within a day or so of my attepmt, they called in The Techs to set them up a VPN (since they'd heard from me that VPNs are a good thing), and now they use Remote Desktop and call it a VPN. True, they can do what I was promising -- they can access their email, files, and FileMaker database remotely -- but it's also damned inefficient, and it runs over a VPN because RDP doesn't do crypto.
Anyway, bit of a rant, but if there's a moral, it's this: If you can't afford a full IT person, you certainly can't afford freelance MSCE dicks.