There should be something like Godwin's Law that says, whenever someone equates a forgetting memory of a single human to large-scale computer-assisted data-mining, he automatically loses and the thread ends.
Yes, I want to be able to lie about a minuscule event if I want to. And there's nothing wrong with that. Forgetting or not knowing is what makes society function. Knowing that not every move of yours is recorded is what keeps you sane.
If you never were in a situation that you were glad was not recorded on photo/video, you are either a fetus with typing skills or a basement-dwelling troll. Honest question my ass.
W already is a "per time" unit. Multiplying it back with time gets you energy, which is what people pay for. Energy price is $/€ per(hah!) kWh. A device draws xW of power, etc.
Never ever do you have a use for W/time in daily life and people using it always mean either W or Wh but don't understand what they're saying.
I'll give you the benefit of the doubt that you're not trolling, although equalling MWh to MW per h is pretty close.
"rate of flow", "volume", is that what they teach in schools nowadays? Watt is the unit of energy conversion per time, also known as work. Watt times time is energy. Watt per time would be energy per time^2 and makes absolutely no sense.
Noone outside of power plants should ever use Watts per time:
These systems only work when you share as much as you receive. Kind of like people who are dead set against Google knowing minor anonymous details about their life yet at the same time are happy for live traffic feedback in navigation apps.
What. The. Fuck.
So for live traffic feedback I should have to provide back my live GPS location or even totally unrelated information? What a completely retarded argument. You just pulled a totally unrelated analogy out of your ass.
ZFS exists in more than stable incarnations if you aren't too stubborn to force Linux on yourself but instead, use the right tool for the job. The problem with Linux users is, they think Linux _is_ the right tool for _every_ job until... they reinvent the proper tool, badly. Rinse and repeat.
Following a path-of-least-astonishment approach is not automatically abandoning the KISS principle of UNIX or its ability to shoot yourself in the foot. First and foremost, a tool should do its job in a non-surprising, intuitive and (hopefully) properly documented way. Any ability to shoot yourself in the foot is just icing on the cake _if the tool does its main job properly_.
Example: rm -fr / _should_ remove my whole filesystem tree, no questions asked, because the path of least astonishment requires that there are no edge cases or unforseen checks coded in. It's the same reason I can edit a directory entry with vi if I want to.
However, if a tool is touted as fully integrity-checking its tree using SHA-1 checksum, I would assume it does this in a consistent manner on every operation. Otherwise, what's the point?
Forcefully leaving the POLA to be able to shoot yourself in the foot is not UNIX. That's Linux.
Git sync to a lot of drives IS A BACKUP. It is exactly what an ideal backup should be, historic, up to date, minimizes storage. What is that system if it isn't an automatic backup!
A backup is not a backup if you overwrite it each time. They weren't doing backups at the file level, they were doing backups at the "Git level".
Mirroring Git repositories is a form of backup, yes. But only if Git itself works flawlessly. A backup is a _disaster_ recovery mechanism. Disaster includes software failing. Using Git to fulfill the "historic" requirement of backups is a problem if Git itself fails.
Looks like they trusted the Git adverts too much and thought complete integrity means complete integrity _on every access_ when in reality, it's nothing more than a ZFS fs that requires scrub to notice any corruption. In other words, one group of half-asses relied on another group of half-asses doing their job properly.
_Always_ double-check and make sure you know how your tools work, especially under disastrous circumstances.
You mean like determining which page to serve based on the client-side information of which page they want?
If you want to base decisions on what URLs you served, that's perfectly fine, since it's based on facts.
Also note how I said serious decision. It's even spelled out in the official spec that the referrer is completely optional and its contents are at the client's discretion. If someone allows referrers to influence his business model, he's pretty stupid in my eyes. Even IP adresses used for geolocation are better data, although also flawed.
So coming here crying not to take his precious referrer away deserves ricidule. Don't like it? Don't use HTTP, invent your own. Problem solved.
So now you run 24GB of non-ECC memory in a file server.
It's OK, everyone is allowed to shoot themselves in the foot as hard as they want. But personally, I wouldn't announce it in a public discussion and make a fool out of myself.
Looks like the problem is you running a "file server" on desktop-class hardware, then complaining that several small bits of fail end up as one big mess.
Seriously, NVidia graphics in a server with binary blobs disguised as drivers and all that? You're doing it wrong. Period.
Problem is, too many people doing it wrong come up with the next generation of ideas to fuck up perfectly working systems even more. See article.
Slashdot Mirror
The only sick fuck I see is you.
There should be something like Godwin's Law that says, whenever someone equates a forgetting memory of a single human to large-scale computer-assisted data-mining, he automatically loses and the thread ends.
Yes, I want to be able to lie about a minuscule event if I want to. And there's nothing wrong with that. Forgetting or not knowing is what makes society function. Knowing that not every move of yours is recorded is what keeps you sane.
If you never were in a situation that you were glad was not recorded on photo/video, you are either a fetus with typing skills or a basement-dwelling troll. Honest question my ass.
W already is a "per time" unit. Multiplying it back with time gets you energy, which is what people pay for. Energy price is $/€ per(hah!) kWh. A device draws xW of power, etc.
Never ever do you have a use for W/time in daily life and people using it always mean either W or Wh but don't understand what they're saying.
I'm sorry. Work is supposed to read rate of work. Yes, it can be confusing.
I'll give you the benefit of the doubt that you're not trolling, although equalling MWh to MW per h is pretty close.
"rate of flow", "volume", is that what they teach in schools nowadays? Watt is the unit of energy conversion per time, also known as work. Watt times time is energy. Watt per time would be energy per time^2 and makes absolutely no sense.
Noone outside of power plants should ever use Watts per time:
http://en.wikipedia.org/wiki/Watt#Confusion_of_watts.2C_watt-hours_and_watts_per_hour
Now get off my lawn you uneducated fool.
Wrong. Wrong. Wrong. If anyone says "Watts per time" again, I personally hunt you down and smack your uneducated hipster face.
It "costs" 41MWh per hour. Period.
We had traffic jam reporting via radio before anyone knew what a smartphone was and before you had always-on Internet whereever you went.
http://en.wikipedia.org/wiki/Traffic_Message_Channel
Do you really want to waste traffic with every individual car uploading its position in realtime? That's even more braindead.
What. The. Fuck.
So for live traffic feedback I should have to provide back my live GPS location or even totally unrelated information? What a completely retarded argument. You just pulled a totally unrelated analogy out of your ass.
You mixed up your wording.
Do it right the first time, optimize for speed later. You don't want to find out you're unable to optimize because the design is flawed.
Yeah, if only it had that, because it hasn't.
Hardlinks. Hardlinks!
Let's tie our backup to the live data so we never know when either gets damaged and, as a bonus, damages the other as well!
Hardlinks are nothing. They are not snapshots and certainly not backups.
ZFS exists in more than stable incarnations if you aren't too stubborn to force Linux on yourself but instead, use the right tool for the job. The problem with Linux users is, they think Linux _is_ the right tool for _every_ job until... they reinvent the proper tool, badly. Rinse and repeat.
Uh, WTF?
Following a path-of-least-astonishment approach is not automatically abandoning the KISS principle of UNIX or its ability to shoot yourself in the foot. First and foremost, a tool should do its job in a non-surprising, intuitive and (hopefully) properly documented way. Any ability to shoot yourself in the foot is just icing on the cake _if the tool does its main job properly_.
Example: rm -fr / _should_ remove my whole filesystem tree, no questions asked, because the path of least astonishment requires that there are no edge cases or unforseen checks coded in. It's the same reason I can edit a directory entry with vi if I want to.
However, if a tool is touted as fully integrity-checking its tree using SHA-1 checksum, I would assume it does this in a consistent manner on every operation. Otherwise, what's the point?
Forcefully leaving the POLA to be able to shoot yourself in the foot is not UNIX. That's Linux.
A backup is not a backup if you overwrite it each time. They weren't doing backups at the file level, they were doing backups at the "Git level".
Mirroring Git repositories is a form of backup, yes. But only if Git itself works flawlessly. A backup is a _disaster_ recovery mechanism. Disaster includes software failing. Using Git to fulfill the "historic" requirement of backups is a problem if Git itself fails.
Looks like they trusted the Git adverts too much and thought complete integrity means complete integrity _on every access_ when in reality, it's nothing more than a ZFS fs that requires scrub to notice any corruption. In other words, one group of half-asses relied on another group of half-asses doing their job properly.
_Always_ double-check and make sure you know how your tools work, especially under disastrous circumstances.
Just because there are weaker links doesn't mean stronger links should stop getting stronger or even become weaker.
No idea WTF the author is smoking. Must be one of those post-privacy pussies, too, who just give up.
If you want to base decisions on what URLs you served, that's perfectly fine, since it's based on facts.
Also note how I said serious decision. It's even spelled out in the official spec that the referrer is completely optional and its contents are at the client's discretion. If someone allows referrers to influence his business model, he's pretty stupid in my eyes. Even IP adresses used for geolocation are better data, although also flawed.
So coming here crying not to take his precious referrer away deserves ricidule. Don't like it? Don't use HTTP, invent your own. Problem solved.
Congratulations. You just realised why any of this marketing/analysing/whatever bullshit has no place in an environment like HTTP.
The referrer is a client-side piece of information. Making any serious decision based on client-side data is foolish to begin with.
HTTP is you serving pages and clients requesting pages. The End.
If your fascist monitoring needs are not served by HTTP, go invent your own protocol and see if anyone cares.
RefControl addon.
Stop making this about page impressions. It's a ridiculous theory by any means.
This is about oil vs. green.
Yeah, those pesky facts and how they bully the liars. Burn facts!
Idiot.
Put it in a closet. If you have visitors, proudly show it to them. Leave it at that.
Running this thing is irresponsible.
I dunno, maybe that the Linux guys have to report to MS to have their stuff working? Duh?
So now you run 24GB of non-ECC memory in a file server.
It's OK, everyone is allowed to shoot themselves in the foot as hard as they want. But personally, I wouldn't announce it in a public discussion and make a fool out of myself.
Your use case has no relevance to the discussion.
No. In the spirit of openness, hopefully this bullshit will get eaten by the anti-monopoly regulation.
Giving in to this bullshit was the most stupid thing the Linux guys could do.
Looks like the problem is you running a "file server" on desktop-class hardware, then complaining that several small bits of fail end up as one big mess.
Seriously, NVidia graphics in a server with binary blobs disguised as drivers and all that? You're doing it wrong. Period.
Problem is, too many people doing it wrong come up with the next generation of ideas to fuck up perfectly working systems even more. See article.