Maybe, but depends on what your metric is. Most mechanical HDs are very good a sequential reads, but performance drops to abysmal values at random reads. So, considering a consumer-grade disk with noticeable fragmentation, 1Gbit may not be that slow. Of course, you can make local storage blow 1Gbit/s out of the water, but I don't see that often Photoshop users with hardware raid setups or server-grade disks. Or SSD.
I have 1Gbit FTTH available where I live, but my internet access is 120Mbit. It's faster than many company's LAN. But internet access speed is only relevant if the provider itself is geographically close, and can cope with the demand. 1Gbit access is useless when accessing the staggering amount of content being served by 100Mbit and even 10Mbit servers, or cheap vps systems.
So you have 100Gbit, you're pulling the data out of what? Thin air? And what about latency, how far are you pulling the data? It doesn't compute.
And 6Gbit isn't the limit, there are other options.
that's like saying that windows doesn't get viruses except when it does. pretty meaningless. when repository packages flag insecure (hey, at least they are flagged), it is usually because the package index is out of date (gosh its hard to type "apt-get update").
Yes, because not only every conceivable software is installed using a repository, but also the repository mantainers can easily verify if a source package not maintained by them has no malicious code in it. It is a nice dream.
you're right, but doesn't that make windows and windows software look even worse? at least there is a contingent of developers and power users (that do have access to the source code) looking after debian repositories. you can make all sorts of rediculous arguments against the security of debian stable repos, but you'll always manage to make windows and associated crapware look much worse for it.
You may call it ridiculous, but it doesn't make them less real. And as far as windows goes, many desktop applications you run on linux are available on windows too. My point is that just because you use repositories, you think you are safe from malicious code or external threats. You're not.
i bet you think you're smart, but i'm actually talking about pentium 4s with 512 Mb RAM or more. windows vista and 7 don't run on these, but the tasks that people normally do in windows that they can't do with this "older" hardware (even web browsing with XP with SP3 is notoriously slow on a pentium 4 if you're also running a virus monitoring program like AVG or Avast in the background).
Why would you install a modern windows version on an old computer? I guess you aren't installing gnome3 or kde 3.7 on them also. I've actually used W7 on a p4 with 1GB and antivirus, and it isn't that bad. I can run many of the open-source desktop applications available for linux AND many of them available for windows. Also, why would I want an old P4 to run linux, if (where I live) the cost of electricity is high enough so an Atom mobo would pay itself by the end of an year?
its funny how often i hear this sort of thing. unfortunately it doesn't tell me anything. do you visit porn sites or software crack sites? have you even connected your windows computer to the internet? my father went years without getting a windows virus, and he knows nothing about computers; and then one day he did and it crippled his system. that you have avoided them so far doesn't mean you won't get infected tomorrow. you can avoid viruses with windows, but with linux you don't have to.
I can't tell you I do, because, you know, liability issues. But hypotetically, let's say yes. A lot. But most modern online threats aren't "viruses", and in that regard stock linux desktop also isn't that safe. The system itself is not easily compromised, but your personal data can be leaked when running a browser with the same uid as you, same as in windows. Do you run your browser as a separate unprivileged user? And how about csrf?
People have different experiences using software. I'm not going to stop using the desktop applications that allow me to be productive just because someone on slashdot thinks I should be running linux. If/when they're available for linux (or BSD) shure, why not? I really don't care if to run a product X I need more ram - ram is dirt cheap, and my lowest spec desktop packs 4GB. I get a (more) uniform desktop experience, a clipboard that actually works with rich media, and the hability to run whatever I want, including many opensource apps. I also get usable vector fonts, more workable area per desktop, generally a better hardware support, and integrated encrypted remote access that actually works with slow connections. Shure, I don't have rdp exposed to the internet, all my systems have dedicated BSD firewalls to unsecure networks. What real advantage would a linux desktop give me over the setup I have now? Can you please elucidate me?
The Road to El Dorado, Antz, Chicken Run, Deep Blue Sea, Star Trek: Insurrection, Fantasia 2000, Men in Black, Hollow Man and many many more, were created with Linux software such as RAYZ, Maya or Shake.
So, your basis is a 2007 article, vague on details. Well, some years had passed, and Shake was discontinued in 2009. And as renderfarms go, mentioning linux is meaningless, because the operating system itself isn't that relevant. You could have mentioned Inferno, Flame and Flare, that are probably the only special effects applications that are first-class citizens in Linux, but usually sold as an appliance. But most software used (Smoke, SoftImage, Maya, 3DS Max, Mudbox, Avid DS, Adobe After Effects, Houdini, etc) either doesn't have a linux version or is a second-class citizen on linux. Most of them require an approved hardware list and only run on RHES, so they're more close to linux-based appliances than to regular desktops. Go ahead, try to install them on your linux desktop. Oddly enough, most of the applications mentioned will run just fine on my windows desktop. But you know what's even funnier? People working with them on linux don't really care if it's linux or not, they use it because it provides the funcionality they need for their work. And every single one of them are closed-source patent-ridden applications, just like the ones you criticized on a previous post.
DraftSight is compatible with AutoCAD dwg, and its developed by Dassault Systems (developers of Catia) so it should definitely be considered "professional grade".
There are a ton of DWG-compatible CAD programs out there. Until they have a bug, or aren't actually compatible with the latest version. And have you tried to hook them up with your plotter? Do you have all the options you need? Because I also know how to make a google search, tnx.
also, many of the 3D animated Hollywood blockbusters are produced
Show me 1 move where the main production/editing/postproduction is being run on linux. It would be a huge surprise, since the-facto standards are Avid and Discreet (now Autodesk) products. I'm not even aware of any special-effects suite for Linux. But yes, I've used CinePaint.
and rendered using Linux
Rendering isn't a desktop job. And saying a given cluster runs linux is meaningless. They also use/used freebsd clusters, Apple XServe clusters, SGI clusters, Sun clusters and many other operating systems. You know why? Because rendering is highly paralelizable, and your custom renderer can be developed in a way that can be run on almost any operating system with network support.
all my linux apps come out of the distro's official secured packagerepository,
Except when they don't (I actually use apps that usually aren't in the official repositories). Or when the repository isn't secure (remember the debian incident?). Or when you don't actually audit the code used to build the package, so you can't be shure the original svn/git/cvs/whatever can't be tainted.
Repositories are just a way of distributing software over the internet/intranet. They make no assumption about quality of code, reliability of the application itself and integrity of the original code. And while you can rest assured in numbers (there were few incidents up until today), I'd suggest it's because it is quite difficult to detect tampering when you actually don't control the source code.
i'm thankful not to have to worry about my machine being bogged down by antivirus and adware programs, having to trawl the web for drivers (such as ac'97) and basic programs like pdf readers (why the hell a pdf reader needs to be more than a couple of hundred kb is beyond me), and having to buy office and graphics packages separately and install them with CDs/DVDs. why is windows always so bloated and slow?
Stop buying crappy hardware. And my Windows is actually faster than any recent linux distro I've tried, for desktop work. But hey, if you don't know how to use it, it's like everything else. Btw, a couple of hundred KB is smaller than the size of the PDF spec itself, so probably a pdf reader that small won't implement all the tiny bits. You don't even need a pdf viewer, you can use ImageMagick/GraphicsMagick and convert the document to jpg pages or png or whatever. Until you need to read a signed or encrypted document, like, you know, legal documents.
i get "old" computers for free because linux runs fine on them, but i do feel sorry for those poor suckers stuck with windows.
That's probably because people buy new computers because they need to do some actual work. I've had a ton of old computers that wouldn't run linux (8088, several 286, several 386, 486, Cyrix (486), Pentium 75/100, etc) . Go ahead, install your favourite distro on a 486DX with 4Mb of ram with X, and tell me how awesome your experience is.
I find it funny that when I tell that the X-based Unix desktop sucks, I allways get those "windows is bloated blahblah" and "just do this and that and use this distro" and "what you really need is this and that". I've been casually trying linux since 1995, so yeah, I have a pretty good idea about what's available, and the pros and cons for my kind of work. Part of my job description is BSD systems administrator (and has been for the last 12 years), so I'm not new to the whole "unix parade". Operating systems are tools, all of them have strengths and weaknesses, and disregarding them just because "whoa I had a virus on my pc so linux is so awesome" is stupid. Btw, I use Windows since v2.6, and the last virus I got was around 1995 - and it was a DOS virus.
I've rewritten the SELECT to return 3 columns, and in 8.4 using IN is a bit faster than using JOIN (but not by a relevant difference), so you are probably right.
The thing is that prior to this release, applications that were using SQL joins (involving more than 2 tables) where ruled OUT:
Now a lot of more applications *can* use MySQL Cluster.
Yeah, that's a less flattering remark than a 70x increase in speed, but probably more useful:D. I'm not familiarized with MySQL Cluster, and I'm surprised it is_that_ different from the server version.
I'm with you on this one. 1.23 seconds seems like a dreadful result, and I can't imagine why they thought it was a good example.
The only heavy MySQL application I use is Magento (and it relies heavily on JOINs), and a small server can blow these results out of the water, on similar-sized tables, and with multiple concurrent queries (but only using 3 to 5 JOINs per query).
Did you test the queries on the actual database server, or did you just run your script? Were you using regular drivers, or PDO? Was the SQL server on the same machine, or on a separate one?
Just because your script didn't run well, doesn't mean that SQL Server was slower. Probably the PHP database modules had a lot to do with it - I guess there is much more attention given to MySQL integration than with SQL Server, or some misconfiguration (always 2-3 seconds? Maybe is a DNS issue, were you using hostnames or just IP's?).
I've just benchmarked 2 simple queries on PostgreSQL 8.2 and 8.4, and in 8.4 the IN clause is being translated to a hash semi-join, so it is faster (at least otn the simple testcase) than using JOIN.(I've already posted the queries, so if interested, scrolll up) While performance is important. using joins on complex queries usually result on an ugly mess, and I usually prefer maintainability over performance. Also, the parent is right - for complex expressions, it is way easier to test the individual subselects than to try to figure out what's going on on a multi-join expression.
Following this thread, I decided to benchmark (with EXPLAIN ANALYZE) a simple IN vs JOIN statement on both PostgreSQL 8.2 and PostgreSQL 8.4.
The statements are something as follows:
SELECT * FROM STATEMENTS WHERE CLIENT IN (SELECT ID FROM CLIENTS WHERE STATUS=1);
SELECT * FROM STATEMENTS JOIN CLIENT ON STATEMENTS.CLIENT=CLIENTS.ID WHERE CLIENTS.STATUS=1;
The statements table has about 2.5 million rows and clients has about 30k rows. The match (status=1) results in roughly 10k rows.
Now the interesting part - In 8.2, using JOIN is quite faster, but in 8.4, using IN is actually faster than using JOIN (at least for these queries), because it uses a hash table and a hash join, insted of a full join.
That's the most unfortunate program name I've heard in a while:D
The biggest advantage of ethertool and such is that is console-based, so easily integrated in scripts and such. Here's an output example of mii-tool:
eth0: negotiated 100baseTx-FD flow-control, link ok
eth1: negotiated 100baseTx-FD flow-control, link ok
No, I don't mean hypothetically, China has stopped buying US bonds. It's allowing its holdings to decline as they mature. Yet the resale price is still high, the interest rates are still low. The market would love China to dump some US bonds at fire-sale prices.
Some, probably yes. All of them, I seriously doubt. And everyone's been looking distracted with Europe, so no one (outside of the US) is really paying attention to what's going on.
Japan holds about 6%, the UK has increased its US holdings to about 4%. There's nothing special about China's ownership of US debt. There's nothing that gives them leverage. And, of course, US institutions, including the Social Security Trust, own about 70% of US Federal debt.
Do you think the Social Security Trust are gold bars stored in some banks? The trust is probably scattered along long-term stable investments, and probably some of them outside the US. How do you think those investments would turn out when faced with a significant depeciation of the currency? And all the private sector, how would it react? You'd probably have hundeds of big companies closing doors. Pension funds would evaporate, as well as every single penny deposited on a US bank.
But, as I said in my previous comment, I was speaking hipoteticaly - I don't believe any of those scenarios is viable, but they are indeed possible.
Panics are hard to maintain. The US govt would begin buying up the highest-interest bonds (as they do anyway). Eventually the more aggressive bond traders would say, "Fuck this, I want in", and start buying up the cheapest high-interest bonds left. Within a couple of days the markets would be racing to absorb the whole tranche before the party stops, and prices would be climbing again. Within weeks, the market will be back where it started as if the whole thing never happened.
(...)
You massively over-estimate the size of China's holdings. US domestic buyers alone could absorb dumped Chinese-owned bonds. Hell, the Social Security Trust Fund alone could absorb dumped Chinese-owned bonds, and if the price was right they would.
Panics don't need to be maintained for much time. On the last decade, several apparently solid financial firms collapsed from night to day. From the top of my head, I think China holds around 1.3 trillion in US bonds. If they flooded the market, the dollar value would drop... a lot. And interest rates would go trough the roof. So, even assuming the SS trust fund could cash out 1.3T from night to day (which I sincerely doubt, as usually the money is scattered across a portfolio of stable investments), it would be dollars - witch would evaluate next to nothing. A country in debt cannot generate instant wealth by itself, that's why there is foreign debt.
A drop in the value of the US dollar would make US exports cheap. Increasing US manufacturing and farming. As well as encouraging import-substitution at home. Within ten years, the US would be a manufacturing powerhouse, and the US dollar would be climbing again. (IMO, the US dollar would be lower right now, except for the clusterfuck in Europe. That is slowing the US recovery.)
The inability for nations like Greece to do this is why they are in the toilet. By sharing the Euro, Germany benefits from a lower value currency, allowing it to export more cheaply, and reducing imports. If they still had the Deutsch-Mark, it would be going up in value, effectively raising German prices and wages, German exports would be more expensive, and foreign imports would be cheaper in Germany. Greece, OTOH, lost its manufacturing base, has high debt in a "foreign" currency, under normal circumstances the Drachma would drop in value, effectively lowering wages/etc and encouraging export development. But because the are locked into the Euro, and Euro debt. they've been drifting further and further out to sea until the GFC hit and precipitated a debt crisis.
A drop in the value of the dollar would increase the price of every US import, including raw production materials. Looking at 10 years, yes, probably the economy would recover, stronger than before, but never to the same levels. And after 6 or 7 years of unspeakable depression. You are also assuming the US wouldn't have competition in their exports, which again would be taking out of equation countries like China, India, Brazil, Russia, Japan, etc.
As an european, I'd say the problem here is the same as in the US - the lack of long-term vision and clear leadership. That given, the Euro has been quite more stable than the dollar when considering year-long averages. Also, you have some facts wrong:
- Greece should never been allowed in the Euro. They cooked their books for years, and the complacency of the regulation mechanism allowed it to happen. They still need a hefty sum of money, but partially because the first rescue wasn't done soon enough. I'd say they are politically unstable - too many relevant extremist parties, but the truth is that, even with what is going on, their economy is growing. Also, their main creditors are Germany and France, so their debt isn't in a foreign currency.
- German currency before the euro was quoted at about 0.5Eur, so the Mark was way lower than the Euro. The benefits of the uniformization of the european marked guaranteed by the Euro covered by far the problem of having a strong c
That's why they've been actively investing both in emerging economies and the european market - to reduce their US dependency (the US is by far the biggest buyer of chinese products). Also, the asian economy is booming and even the chinese domestic market is growing at a very rapid pace. What is a fact today may not be tomorrow, and I believe we'll see a global shift in power in the next years.
If China did dump all their investments without any regard for financial losses, very few would consider them "free bonds", and more like "a risky product", as they'd be suspicious the chinese investors knew something they didn't. The markets aren't driven by logic, but by risk assessment and rumors. Also, most 1st world countries have liquidity problems right now, so probably they couldn't even sell them all at once.
If they wanted to get rid of it, they could do it quietly over a period of time and probably even using other nations as proxy. Prices would go down, but it could be done if they're willing to sacrifice their profit.
Of course today this is purely speculation, because US bonds are "equivalent to gold" as a loan collateral. Dumping them would increase interest rates of the loans China has in other countries, so it wouldn't make them any good.
A gradual shift from the dollar to another international trading currency would cause far more damage than dumping all those US titles. Given that the US has little to no control over their currency abroad, and some rumors that upto 40% of all american currency in circulation is fake, the world could face an unrecoverable devaluation of the dollar, with devastating effects to many countries.
I'm addressing the incessant refrain that China somehow can "Call in" US debt, as if they were regular loans. There's no mechanism by which China can use US Treasury issued bonds to harm the US. There's no mechanism by which they can even threaten it. There is no leverage. People need to stop uncritically regurgitating this hysterical nonsense.
There are several hypothetical ways to make that happen. If China stopped buying US bonds, not only that would have an effect on short-term liquidity, but probably would have a cascading effect, and very few countries have China's buying power. Given the valorization of the yuan relative to the dollar, and considering the effort China is making to free themselves from the US dependency, I'd guess they will shift from usd to another international currency sometime in the next years, and that will not only affect dollar quotation, but also US imports from China, which can lead to an increased cost of living. Another way is if China started to dump quietly their US bonds. A direct effect would be an increase in loan rates due to the devalorization of the bonds themselves and increased risk perception - assuming anyone would buy them. If the chinese started to sell their US bonds, many analysts would become worried, specially given the history of corporate espionage.
Well, I know if I were writing a 10k+ page document, I'd want to show a little pride in my work.
The pride should be the same regardless of document size. Try to do wyswig editing of a document that big with database integration or external resources on Linux. You can proclaim LaTex superior typesetting, but that doesn't mean it's "correct" (as an example, most kerning corrections are beautifiers and are usually applied to incorrectly designed fonts. Either that or every document composed up until the late eighties is wrong - including LaTex documents as well). But hey, I could also create the document in postscript with a text editor. Or create my own formatting language (it wouldn't be the first time).
At the expense of learning a completely new system, which while may do what the other program doesn't, itself be unusable for tasks you typically used the other one for. The cost in fixing the bug with the first in time can often be less than the time to find and learn the new program.
What completely new system? Most programs mimicks the most used interfaces and shortcuts! Even java applications don't look that bad. If you can pick up a 30-40Mb source tree that you know nothing of, and debug and correct a problem that may imply field-specific knowledge all by yourself faster than a person can pick a familiar interface, you must be some sort of genius. And then, go ahead, try to submit the patch to the project.
Give me a choice of five programs that between them do what I want for a specific task, but are closed and can die at any time, and a single one that does everything but a minor issue which I myself can fix, and I'll invest the time that would have been spent with the five programs fixing the one, and then everyone benefits from your use of time.
The biggest fallacy is that you usually don't have a "single one that does everything" in oss. Not even "one that does most of things". You usually have 3 or 4 competing, uncomplete and often bug-ridden applications to do the same basic tasks, and none that provides advanced tasks. And with field-specific applications, you're often way better off with a commercial product than with a community "hack" (with all respect for those who dedicated their time to develop). Just because it works and produces it doesn't mean it works well, or does its function correctly.
In userland, as you seem to agree, it's just files. And mount is a userland tool. So mount should deal with files (not devices described in some other way), as far as the user is concerned.
As I said, raw block devices aren't "plain old" files, but are accessed on a file-like manner. While mount is an external tool, it is a syscall wrapper, so the actual mounting/unmounting is done by the kernel, and not by "mount".
If your file is a block device, yes, mount can handle it. If your file is a plain binary file (and you may have thousands of those on your computer), mount doesn't have to recognize it. Following the same philosophy, the "cd" command won't show you the contents of a text file, or fsck won't correct the xml structure of a document file.
If you are using tested server scripts* without trying to do anything tricky can be as secure as anyone else. For example if you just wanted a FTP server you could set up a anon chrooted vsftpd server, secure ssh and set up a firewall to drop all other connections (these are well tested FOSS "industry" standard software that take no time to set up if you have done it before).
I did not fully understood what you're trying to say, but I can tell you this - there's no way you can secure a VPS if their infrastructure was hacked and storage systems were exposed. And in many countries, you are legally responsible for any crime carried out using your VPS. So yes, I do care.
My audio subsystem has been running with 6msec latency for the last three years, so I'd pretty much consider it fixed..
With a stock kernel? Or did you have to fiddle with the system until you got a working configuration?
When you are writing with that many pages, you are in general writing something that will become a rather large book or reference material, typesetting matters (it does so even in small documents too). MS word is horrible at typesetting (not as bad as it used to be, but still pretty nasty) and is not used to typeset books for a reason.
I'm sorry to disapoint you, but many publishers DO use MS Word. I do know Latex is the best choice for many scientific (with formulas/equations) documents and very popular in some fields, but it is by far a first choice to produce print-ready books. Also, it is very easy to export a Word document to a DTP publishing package with more advanced typesetting. As an extra, I can generate db-driven documents, such as mail notifications, that don't require that much typesetting precision, and easily train someone to do it.
In hindsight I was thinking of another linux video related item and not cinelerra, so I'll concede on this one. Last time I used it was in 2005 or so, and while I agree it was pretty nasty back then I had assumed it had progressed in some way. I'd imagine lightworks would fit the bill though regardless.
Yes, Linux is probably the first choice for rendering farms, storage/capture systems and even some custom developed modelling and editing software. But most modelling software, special effects and non-linear editing runs on OSX or Windows workstations.
Well why are you using word to try to make large high quality documents? even with any amount of hacking it cannot be made suitable for the purpose. My biggest problem with proprietary software is when you hit a problem it was not originally designed to handle. If software is designed as a set of discrete parts you can mix them in certain ways or add your own to achieve what you wish, and this is how oss stuff is made and evolves. When it is all an integrated untouchable lump this becomes an issue.
You assumed I was doing high-quality documents, I never told you that. And I can use Latex in other operating systems - in fact, Latex isn't even a "linux technology", as it can be run in virtually any unix operating system and Windows. Most programs you mentioned are "the only choice" for many kinds of work - in Windows you have a complete software ecosystem. If some program does have a bug, you can use a similar tool from other provider. That happened to me last week when Gimp rendered incorrectly a medium-size uncompressed TIFF - I used another program to do the editing I needed.
OSS isn't better at handling unexpected problems than proprietary software. Having the source and a community only helps you if you have time to spend and the knowledge to fix it. If I need to do X, and there's a very good program to do X, I really don't care if it's oss or paid software, if it is the best tool for the job. I have no problem in using Linux if it is the right tool for the job, but for GUI desktop work, it rarely is. I work with OSS operating systems everyday (a bit of Linux, a lot of BSD), and I work with dozens of oss applications that are very good at what they do. When I develop software, I have a tendency to choose open environments and use oss libraries, if they are the right choice. But oss isn't the right choice everytime, and touting it as the ultimate solution is stupid.
If you have those documents on paper, probably a scanned copy will hold no legal value. You should check if your local legislation allows you to present digital copies instead of the originals if needed. If you work at home or have a small business, you should check the accounting procedures for electronic documents, as many countries have different ideas of what an "electronic document" is (and most of them won't allow you to just digitalize the documents and throw the originals away).
Storing paper isn't that much of an hassle, when compared to long-term digital storage, specially if you don't plan to need it soon. Buy a box of binders and separate the documents by year/month, or create an index.
Slashdot Mirror
Maybe, but depends on what your metric is. Most mechanical HDs are very good a sequential reads, but performance drops to abysmal values at random reads. So, considering a consumer-grade disk with noticeable fragmentation, 1Gbit may not be that slow. Of course, you can make local storage blow 1Gbit/s out of the water, but I don't see that often Photoshop users with hardware raid setups or server-grade disks. Or SSD.
I have 1Gbit FTTH available where I live, but my internet access is 120Mbit. It's faster than many company's LAN. But internet access speed is only relevant if the provider itself is geographically close, and can cope with the demand. 1Gbit access is useless when accessing the staggering amount of content being served by 100Mbit and even 10Mbit servers, or cheap vps systems.
So you have 100Gbit, you're pulling the data out of what? Thin air? And what about latency, how far are you pulling the data? It doesn't compute.
And 6Gbit isn't the limit, there are other options.
Do you have a SSD disk? Photoshop almost looks like a regular application when opening... And lightroom is actually usable.
that's like saying that windows doesn't get viruses except when it does. pretty meaningless. when repository packages flag insecure (hey, at least they are flagged), it is usually because the package index is out of date (gosh its hard to type "apt-get update").
Yes, because not only every conceivable software is installed using a repository, but also the repository mantainers can easily verify if a source package not maintained by them has no malicious code in it. It is a nice dream.
you're right, but doesn't that make windows and windows software look even worse? at least there is a contingent of developers and power users (that do have access to the source code) looking after debian repositories. you can make all sorts of rediculous arguments against the security of debian stable repos, but you'll always manage to make windows and associated crapware look much worse for it.
You may call it ridiculous, but it doesn't make them less real. And as far as windows goes, many desktop applications you run on linux are available on windows too. My point is that just because you use repositories, you think you are safe from malicious code or external threats. You're not.
i bet you think you're smart, but i'm actually talking about pentium 4s with 512 Mb RAM or more. windows vista and 7 don't run on these, but the tasks that people normally do in windows that they can't do with this "older" hardware (even web browsing with XP with SP3 is notoriously slow on a pentium 4 if you're also running a virus monitoring program like AVG or Avast in the background).
Why would you install a modern windows version on an old computer? I guess you aren't installing gnome3 or kde 3.7 on them also. I've actually used W7 on a p4 with 1GB and antivirus, and it isn't that bad. I can run many of the open-source desktop applications available for linux AND many of them available for windows. Also, why would I want an old P4 to run linux, if (where I live) the cost of electricity is high enough so an Atom mobo would pay itself by the end of an year?
its funny how often i hear this sort of thing. unfortunately it doesn't tell me anything. do you visit porn sites or software crack sites? have you even connected your windows computer to the internet? my father went years without getting a windows virus, and he knows nothing about computers; and then one day he did and it crippled his system. that you have avoided them so far doesn't mean you won't get infected tomorrow. you can avoid viruses with windows, but with linux you don't have to.
I can't tell you I do, because, you know, liability issues. But hypotetically, let's say yes. A lot. But most modern online threats aren't "viruses", and in that regard stock linux desktop also isn't that safe. The system itself is not easily compromised, but your personal data can be leaked when running a browser with the same uid as you, same as in windows. Do you run your browser as a separate unprivileged user? And how about csrf?
People have different experiences using software. I'm not going to stop using the desktop applications that allow me to be productive just because someone on slashdot thinks I should be running linux. If/when they're available for linux (or BSD) shure, why not? I really don't care if to run a product X I need more ram - ram is dirt cheap, and my lowest spec desktop packs 4GB. I get a (more) uniform desktop experience, a clipboard that actually works with rich media, and the hability to run whatever I want, including many opensource apps. I also get usable vector fonts, more workable area per desktop, generally a better hardware support, and integrated encrypted remote access that actually works with slow connections. Shure, I don't have rdp exposed to the internet, all my systems have dedicated BSD firewalls to unsecure networks. What real advantage would a linux desktop give me over the setup I have now? Can you please elucidate me?
The Road to El Dorado, Antz, Chicken Run, Deep Blue Sea, Star Trek: Insurrection, Fantasia 2000, Men in Black, Hollow Man and many many more, were created with Linux software such as RAYZ, Maya or Shake.
So, your basis is a 2007 article, vague on details. Well, some years had passed, and Shake was discontinued in 2009. And as renderfarms go, mentioning linux is meaningless, because the operating system itself isn't that relevant. You could have mentioned Inferno, Flame and Flare, that are probably the only special effects applications that are first-class citizens in Linux, but usually sold as an appliance. But most software used (Smoke, SoftImage, Maya, 3DS Max, Mudbox, Avid DS, Adobe After Effects, Houdini, etc) either doesn't have a linux version or is a second-class citizen on linux. Most of them require an approved hardware list and only run on RHES, so they're more close to linux-based appliances than to regular desktops. Go ahead, try to install them on your linux desktop. Oddly enough, most of the applications mentioned will run just fine on my windows desktop. But you know what's even funnier? People working with them on linux don't really care if it's linux or not, they use it because it provides the funcionality they need for their work. And every single one of them are closed-source patent-ridden applications, just like the ones you criticized on a previous post.
DraftSight is compatible with AutoCAD dwg, and its developed by Dassault Systems (developers of Catia) so it should definitely be considered "professional grade".
There are a ton of DWG-compatible CAD programs out there. Until they have a bug, or aren't actually compatible with the latest version. And have you tried to hook them up with your plotter? Do you have all the options you need? Because I also know how to make a google search, tnx.
also, many of the 3D animated Hollywood blockbusters are produced
Show me 1 move where the main production/editing/postproduction is being run on linux. It would be a huge surprise, since the-facto standards are Avid and Discreet (now Autodesk) products. I'm not even aware of any special-effects suite for Linux. But yes, I've used CinePaint.
and rendered using Linux
Rendering isn't a desktop job. And saying a given cluster runs linux is meaningless. They also use/used freebsd clusters, Apple XServe clusters, SGI clusters, Sun clusters and many other operating systems. You know why? Because rendering is highly paralelizable, and your custom renderer can be developed in a way that can be run on almost any operating system with network support.
all my linux apps come out of the distro's official secured packagerepository,
Except when they don't (I actually use apps that usually aren't in the official repositories). Or when the repository isn't secure (remember the debian incident?). Or when you don't actually audit the code used to build the package, so you can't be shure the original svn/git/cvs/whatever can't be tainted.
Repositories are just a way of distributing software over the internet/intranet. They make no assumption about quality of code, reliability of the application itself and integrity of the original code. And while you can rest assured in numbers (there were few incidents up until today), I'd suggest it's because it is quite difficult to detect tampering when you actually don't control the source code.
i'm thankful not to have to worry about my machine being bogged down by antivirus and adware programs, having to trawl the web for drivers (such as ac'97) and basic programs like pdf readers (why the hell a pdf reader needs to be more than a couple of hundred kb is beyond me), and having to buy office and graphics packages separately and install them with CDs/DVDs. why is windows always so bloated and slow?
Stop buying crappy hardware. And my Windows is actually faster than any recent linux distro I've tried, for desktop work. But hey, if you don't know how to use it, it's like everything else. Btw, a couple of hundred KB is smaller than the size of the PDF spec itself, so probably a pdf reader that small won't implement all the tiny bits. You don't even need a pdf viewer, you can use ImageMagick/GraphicsMagick and convert the document to jpg pages or png or whatever. Until you need to read a signed or encrypted document, like, you know, legal documents.
i get "old" computers for free because linux runs fine on them, but i do feel sorry for those poor suckers stuck with windows.
That's probably because people buy new computers because they need to do some actual work. I've had a ton of old computers that wouldn't run linux (8088, several 286, several 386, 486, Cyrix (486), Pentium 75/100, etc) . Go ahead, install your favourite distro on a 486DX with 4Mb of ram with X, and tell me how awesome your experience is.
I find it funny that when I tell that the X-based Unix desktop sucks, I allways get those "windows is bloated blahblah" and "just do this and that and use this distro" and "what you really need is this and that". I've been casually trying linux since 1995, so yeah, I have a pretty good idea about what's available, and the pros and cons for my kind of work. Part of my job description is BSD systems administrator (and has been for the last 12 years), so I'm not new to the whole "unix parade". Operating systems are tools, all of them have strengths and weaknesses, and disregarding them just because "whoa I had a virus on my pc so linux is so awesome" is stupid. Btw, I use Windows since v2.6, and the last virus I got was around 1995 - and it was a DOS virus.
I've rewritten the SELECT to return 3 columns, and in 8.4 using IN is a bit faster than using JOIN (but not by a relevant difference), so you are probably right.
The thing is that prior to this release, applications that were using SQL joins (involving more than 2 tables) where ruled OUT:
Now a lot of more applications *can* use MySQL Cluster.
Yeah, that's a less flattering remark than a 70x increase in speed, but probably more useful :D. I'm not familiarized with MySQL Cluster, and I'm surprised it is_that_ different from the server version.
I'm with you on this one. 1.23 seconds seems like a dreadful result, and I can't imagine why they thought it was a good example.
The only heavy MySQL application I use is Magento (and it relies heavily on JOINs), and a small server can blow these results out of the water, on similar-sized tables, and with multiple concurrent queries (but only using 3 to 5 JOINs per query).
Did you test the queries on the actual database server, or did you just run your script? Were you using regular drivers, or PDO? Was the SQL server on the same machine, or on a separate one?
Just because your script didn't run well, doesn't mean that SQL Server was slower. Probably the PHP database modules had a lot to do with it - I guess there is much more attention given to MySQL integration than with SQL Server, or some misconfiguration (always 2-3 seconds? Maybe is a DNS issue, were you using hostnames or just IP's?).
I've just benchmarked 2 simple queries on PostgreSQL 8.2 and 8.4, and in 8.4 the IN clause is being translated to a hash semi-join, so it is faster (at least otn the simple testcase) than using JOIN.(I've already posted the queries, so if interested, scrolll up) While performance is important. using joins on complex queries usually result on an ugly mess, and I usually prefer maintainability over performance. Also, the parent is right - for complex expressions, it is way easier to test the individual subselects than to try to figure out what's going on on a multi-join expression.
Following this thread, I decided to benchmark (with EXPLAIN ANALYZE) a simple IN vs JOIN statement on both PostgreSQL 8.2 and PostgreSQL 8.4. The statements are something as follows:
SELECT * FROM STATEMENTS WHERE CLIENT IN (SELECT ID FROM CLIENTS WHERE STATUS=1);
SELECT * FROM STATEMENTS JOIN CLIENT ON STATEMENTS.CLIENT=CLIENTS.ID WHERE CLIENTS.STATUS=1;
The statements table has about 2.5 million rows and clients has about 30k rows. The match (status=1) results in roughly 10k rows.
Now the interesting part - In 8.2, using JOIN is quite faster, but in 8.4, using IN is actually faster than using JOIN (at least for these queries), because it uses a hash table and a hash join, insted of a full join.
That's the most unfortunate program name I've heard in a while :D
The biggest advantage of ethertool and such is that is console-based, so easily integrated in scripts and such. Here's an output example of mii-tool:
eth0: negotiated 100baseTx-FD flow-control, link ok eth1: negotiated 100baseTx-FD flow-control, link ok
No, I don't mean hypothetically, China has stopped buying US bonds. It's allowing its holdings to decline as they mature. Yet the resale price is still high, the interest rates are still low. The market would love China to dump some US bonds at fire-sale prices.
Some, probably yes. All of them, I seriously doubt. And everyone's been looking distracted with Europe, so no one (outside of the US) is really paying attention to what's going on.
Japan holds about 6%, the UK has increased its US holdings to about 4%. There's nothing special about China's ownership of US debt. There's nothing that gives them leverage. And, of course, US institutions, including the Social Security Trust, own about 70% of US Federal debt.
Do you think the Social Security Trust are gold bars stored in some banks? The trust is probably scattered along long-term stable investments, and probably some of them outside the US. How do you think those investments would turn out when faced with a significant depeciation of the currency? And all the private sector, how would it react? You'd probably have hundeds of big companies closing doors. Pension funds would evaporate, as well as every single penny deposited on a US bank.
But, as I said in my previous comment, I was speaking hipoteticaly - I don't believe any of those scenarios is viable, but they are indeed possible.
Panics are hard to maintain. The US govt would begin buying up the highest-interest bonds (as they do anyway). Eventually the more aggressive bond traders would say, "Fuck this, I want in", and start buying up the cheapest high-interest bonds left. Within a couple of days the markets would be racing to absorb the whole tranche before the party stops, and prices would be climbing again. Within weeks, the market will be back where it started as if the whole thing never happened. (...) You massively over-estimate the size of China's holdings. US domestic buyers alone could absorb dumped Chinese-owned bonds. Hell, the Social Security Trust Fund alone could absorb dumped Chinese-owned bonds, and if the price was right they would.
Panics don't need to be maintained for much time. On the last decade, several apparently solid financial firms collapsed from night to day. From the top of my head, I think China holds around 1.3 trillion in US bonds. If they flooded the market, the dollar value would drop... a lot. And interest rates would go trough the roof. So, even assuming the SS trust fund could cash out 1.3T from night to day (which I sincerely doubt, as usually the money is scattered across a portfolio of stable investments), it would be dollars - witch would evaluate next to nothing. A country in debt cannot generate instant wealth by itself, that's why there is foreign debt.
A drop in the value of the US dollar would make US exports cheap. Increasing US manufacturing and farming. As well as encouraging import-substitution at home. Within ten years, the US would be a manufacturing powerhouse, and the US dollar would be climbing again. (IMO, the US dollar would be lower right now, except for the clusterfuck in Europe. That is slowing the US recovery.)
The inability for nations like Greece to do this is why they are in the toilet. By sharing the Euro, Germany benefits from a lower value currency, allowing it to export more cheaply, and reducing imports. If they still had the Deutsch-Mark, it would be going up in value, effectively raising German prices and wages, German exports would be more expensive, and foreign imports would be cheaper in Germany. Greece, OTOH, lost its manufacturing base, has high debt in a "foreign" currency, under normal circumstances the Drachma would drop in value, effectively lowering wages/etc and encouraging export development. But because the are locked into the Euro, and Euro debt. they've been drifting further and further out to sea until the GFC hit and precipitated a debt crisis.
A drop in the value of the dollar would increase the price of every US import, including raw production materials. Looking at 10 years, yes, probably the economy would recover, stronger than before, but never to the same levels. And after 6 or 7 years of unspeakable depression. You are also assuming the US wouldn't have competition in their exports, which again would be taking out of equation countries like China, India, Brazil, Russia, Japan, etc.
As an european, I'd say the problem here is the same as in the US - the lack of long-term vision and clear leadership. That given, the Euro has been quite more stable than the dollar when considering year-long averages. Also, you have some facts wrong:
- Greece should never been allowed in the Euro. They cooked their books for years, and the complacency of the regulation mechanism allowed it to happen. They still need a hefty sum of money, but partially because the first rescue wasn't done soon enough. I'd say they are politically unstable - too many relevant extremist parties, but the truth is that, even with what is going on, their economy is growing. Also, their main creditors are Germany and France, so their debt isn't in a foreign currency.
- German currency before the euro was quoted at about 0.5Eur, so the Mark was way lower than the Euro. The benefits of the uniformization of the european marked guaranteed by the Euro covered by far the problem of having a strong c
That's why they've been actively investing both in emerging economies and the european market - to reduce their US dependency (the US is by far the biggest buyer of chinese products). Also, the asian economy is booming and even the chinese domestic market is growing at a very rapid pace. What is a fact today may not be tomorrow, and I believe we'll see a global shift in power in the next years.
If China did dump all their investments without any regard for financial losses, very few would consider them "free bonds", and more like "a risky product", as they'd be suspicious the chinese investors knew something they didn't. The markets aren't driven by logic, but by risk assessment and rumors. Also, most 1st world countries have liquidity problems right now, so probably they couldn't even sell them all at once.
If they wanted to get rid of it, they could do it quietly over a period of time and probably even using other nations as proxy. Prices would go down, but it could be done if they're willing to sacrifice their profit.
Of course today this is purely speculation, because US bonds are "equivalent to gold" as a loan collateral. Dumping them would increase interest rates of the loans China has in other countries, so it wouldn't make them any good.
A gradual shift from the dollar to another international trading currency would cause far more damage than dumping all those US titles. Given that the US has little to no control over their currency abroad, and some rumors that upto 40% of all american currency in circulation is fake, the world could face an unrecoverable devaluation of the dollar, with devastating effects to many countries.
I'm addressing the incessant refrain that China somehow can "Call in" US debt, as if they were regular loans. There's no mechanism by which China can use US Treasury issued bonds to harm the US. There's no mechanism by which they can even threaten it. There is no leverage. People need to stop uncritically regurgitating this hysterical nonsense.
There are several hypothetical ways to make that happen. If China stopped buying US bonds, not only that would have an effect on short-term liquidity, but probably would have a cascading effect, and very few countries have China's buying power. Given the valorization of the yuan relative to the dollar, and considering the effort China is making to free themselves from the US dependency, I'd guess they will shift from usd to another international currency sometime in the next years, and that will not only affect dollar quotation, but also US imports from China, which can lead to an increased cost of living. Another way is if China started to dump quietly their US bonds. A direct effect would be an increase in loan rates due to the devalorization of the bonds themselves and increased risk perception - assuming anyone would buy them. If the chinese started to sell their US bonds, many analysts would become worried, specially given the history of corporate espionage.
Well, I know if I were writing a 10k+ page document, I'd want to show a little pride in my work.
The pride should be the same regardless of document size. Try to do wyswig editing of a document that big with database integration or external resources on Linux. You can proclaim LaTex superior typesetting, but that doesn't mean it's "correct" (as an example, most kerning corrections are beautifiers and are usually applied to incorrectly designed fonts. Either that or every document composed up until the late eighties is wrong - including LaTex documents as well). But hey, I could also create the document in postscript with a text editor. Or create my own formatting language (it wouldn't be the first time).
At the expense of learning a completely new system, which while may do what the other program doesn't, itself be unusable for tasks you typically used the other one for. The cost in fixing the bug with the first in time can often be less than the time to find and learn the new program.
What completely new system? Most programs mimicks the most used interfaces and shortcuts! Even java applications don't look that bad. If you can pick up a 30-40Mb source tree that you know nothing of, and debug and correct a problem that may imply field-specific knowledge all by yourself faster than a person can pick a familiar interface, you must be some sort of genius. And then, go ahead, try to submit the patch to the project.
Give me a choice of five programs that between them do what I want for a specific task, but are closed and can die at any time, and a single one that does everything but a minor issue which I myself can fix, and I'll invest the time that would have been spent with the five programs fixing the one, and then everyone benefits from your use of time.
The biggest fallacy is that you usually don't have a "single one that does everything" in oss. Not even "one that does most of things". You usually have 3 or 4 competing, uncomplete and often bug-ridden applications to do the same basic tasks, and none that provides advanced tasks. And with field-specific applications, you're often way better off with a commercial product than with a community "hack" (with all respect for those who dedicated their time to develop). Just because it works and produces it doesn't mean it works well, or does its function correctly.
In userland, as you seem to agree, it's just files. And mount is a userland tool. So mount should deal with files (not devices described in some other way), as far as the user is concerned.
As I said, raw block devices aren't "plain old" files, but are accessed on a file-like manner. While mount is an external tool, it is a syscall wrapper, so the actual mounting/unmounting is done by the kernel, and not by "mount".
If your file is a block device, yes, mount can handle it. If your file is a plain binary file (and you may have thousands of those on your computer), mount doesn't have to recognize it. Following the same philosophy, the "cd" command won't show you the contents of a text file, or fsck won't correct the xml structure of a document file.
Why does security matter?
For you, apparently, it doesn't.
If you are using tested server scripts* without trying to do anything tricky can be as secure as anyone else. For example if you just wanted a FTP server you could set up a anon chrooted vsftpd server, secure ssh and set up a firewall to drop all other connections (these are well tested FOSS "industry" standard software that take no time to set up if you have done it before).
I did not fully understood what you're trying to say, but I can tell you this - there's no way you can secure a VPS if their infrastructure was hacked and storage systems were exposed. And in many countries, you are legally responsible for any crime carried out using your VPS. So yes, I do care.
My audio subsystem has been running with 6msec latency for the last three years, so I'd pretty much consider it fixed..
With a stock kernel? Or did you have to fiddle with the system until you got a working configuration?
When you are writing with that many pages, you are in general writing something that will become a rather large book or reference material, typesetting matters (it does so even in small documents too). MS word is horrible at typesetting (not as bad as it used to be, but still pretty nasty) and is not used to typeset books for a reason.
I'm sorry to disapoint you, but many publishers DO use MS Word. I do know Latex is the best choice for many scientific (with formulas/equations) documents and very popular in some fields, but it is by far a first choice to produce print-ready books. Also, it is very easy to export a Word document to a DTP publishing package with more advanced typesetting. As an extra, I can generate db-driven documents, such as mail notifications, that don't require that much typesetting precision, and easily train someone to do it.
In hindsight I was thinking of another linux video related item and not cinelerra, so I'll concede on this one. Last time I used it was in 2005 or so, and while I agree it was pretty nasty back then I had assumed it had progressed in some way. I'd imagine lightworks would fit the bill though regardless.
Yes, Linux is probably the first choice for rendering farms, storage/capture systems and even some custom developed modelling and editing software. But most modelling software, special effects and non-linear editing runs on OSX or Windows workstations.
Well why are you using word to try to make large high quality documents? even with any amount of hacking it cannot be made suitable for the purpose. My biggest problem with proprietary software is when you hit a problem it was not originally designed to handle. If software is designed as a set of discrete parts you can mix them in certain ways or add your own to achieve what you wish, and this is how oss stuff is made and evolves. When it is all an integrated untouchable lump this becomes an issue.
You assumed I was doing high-quality documents, I never told you that. And I can use Latex in other operating systems - in fact, Latex isn't even a "linux technology", as it can be run in virtually any unix operating system and Windows. Most programs you mentioned are "the only choice" for many kinds of work - in Windows you have a complete software ecosystem. If some program does have a bug, you can use a similar tool from other provider. That happened to me last week when Gimp rendered incorrectly a medium-size uncompressed TIFF - I used another program to do the editing I needed.
OSS isn't better at handling unexpected problems than proprietary software. Having the source and a community only helps you if you have time to spend and the knowledge to fix it. If I need to do X, and there's a very good program to do X, I really don't care if it's oss or paid software, if it is the best tool for the job. I have no problem in using Linux if it is the right tool for the job, but for GUI desktop work, it rarely is. I work with OSS operating systems everyday (a bit of Linux, a lot of BSD), and I work with dozens of oss applications that are very good at what they do. When I develop software, I have a tendency to choose open environments and use oss libraries, if they are the right choice. But oss isn't the right choice everytime, and touting it as the ultimate solution is stupid.
If you have those documents on paper, probably a scanned copy will hold no legal value. You should check if your local legislation allows you to present digital copies instead of the originals if needed. If you work at home or have a small business, you should check the accounting procedures for electronic documents, as many countries have different ideas of what an "electronic document" is (and most of them won't allow you to just digitalize the documents and throw the originals away).
Storing paper isn't that much of an hassle, when compared to long-term digital storage, specially if you don't plan to need it soon. Buy a box of binders and separate the documents by year/month, or create an index.