...and so, in other words the switch had nothing to do with performance or stability, it was all about the presence of a much more developed and production-ready.NET toolchain and dev environment. That may be a valid argument for choosing win32 over linux, but it's not the reason the original poster thought it was ("oh it's so much more stable.")
Package versions are an unfair thing to pick on in that respect. The package version has to denote, among other things:
- The upstream baseline version of the source code - Any 3rd party feature or bugfix patches applied - Allow for version bumps when minor packaging issues arise (e.g. adding a "depends" or "requires")
While I agree that in some cases it does get out of hand, the vast majority of that is for packaging reasons where you need a unique version number to represent every package, even if the only thing that has changed is moving around the location of some man file or adding a README or something, or other minor tweaks that relate only to packaging issues and not actual upstream changes.
You may be able to write decent interfaces, or at least you think they are.
But the vast majority of flash-based sites that I have encountered make life harder. Several examples are:
- Tiny 6 point miniscule fixed fonts that I can't resize, contrary to what I've told my browser to use. - Scrollbars that don't obey the mouse scroll whell and often times are not even real scrollbars (i.e. you HAVE to use the up/down buttons, you can't even drag the elevator - sometimes you can't even CLICK on the up/down buttons, you have to hover over them and WAIT for the text to scroll) - Packing the text into a miniscule text area so that the presentation will work on granny's 640x480 display, meaning that to read more than a sentence or two I have to use some godawful nonstandard scrollbar, while the vast majority of my screen space is blank. - Breaking the ability to easily bookmark a section of a page. - Breaking the ability to use the back-button or open-link-in-new-tab features, which I use extensively.
If you call that better than what stock html gives me, then I'm afraid we disagree on what a user interface should be.
If you are one of the flash developers out there that have a clue and don't foister the above nonsense on your web visitors, then all the more power to you. But realize that this "hatred for flash" that you seem to have noticed is because the vast majority of people making flash DO violate one or all of the above annoyances. If that's the case then your beef is with these incompetants soiling the reputation of flash, and not for the rest of us that have been burned by it too many times and have dismissed it.
Microsoft used the BSD networking code in the kernel for its first version of Windows NT. Starting with NT v3.5 in 1994, the TCP/IP stack was completely rewritten from scratch by Microsoft using no BSD code. There is still some BSD code in use in the userspace utilities, like ping, telnet, et cetera.
Can we PLEASE put to rest this notion that any version of Windows made in the last 10 years has any kind of BSD TCP/IP stack code in it?
Umm, on most compilers "int" is a 32 bit value. I don't know what it is on yours. But a 32-bit integer can not come anywhere close to representing 1e12, so essentially what you were asking for was an infinite loop, or whatever loop results from casting (float)1e12 into an integer, which is probably undefined and thus the compiler can't optimise there.
You think the spyware EULAs are bad, try this one for a popular torrent site: 18,554 words or 33 printed pages at 12 pt font and standard margins.
To be fair it's not their EULA, but the one that the original torrentbits web site came up with. Since TB shared its php site source code there are a lot of torrent sites that use a similar or somewhat modified site backend, including the monster EULA.
I think they did it rather as a joke, to make sure that any enforcement agencies reading it would at least have to pay for a couple of extra lawyer-hours of time if they hired councel.
I prefer WinRAR's interface to Winzip, so I install WinRAR and have it handle all compressed archive types (.zip,.rar,.arc, etc.) I'm sure I am not alone in this habit, as WinRAR is extremely popular. There are other programs as well (7-zip, Power Archiver) that handle all the common compression formats and are usually installed in a "use for everything" manner.
My point is that even if you have never encoutered a RAR file previously, there's a chance that your current archiver supports it. Or alternatively put, "lots of people don't use Winzip."/m avoids temptation to take a dig at the crapulence and outdated-ness of winzip.
And car dealerships always love cash because that means all the money from the sale is going to them and not partly to a bank over a period of 5 years (or at least that's my theory, although I do know for a fact that they love cash.)
They may like motivated buyers (i.e. those with cash in hand) but they most certainly don't like missing out on arranging the financing. They have deals with the banks with which they're affiliated, so when you walk in there and arrange financing with them they get a cut of the deal. They make a very large portion of their total profits off these arrangements with lenders and banks, so they most definitely get shafted if you pay cash (or get a loan from your own bank/credit union before walking into the dealership for that matter.)
Why would a mountaineer care if someone sets a record for highest mountain climbed? Why would an explorer care if someone pioneered a part of the antartic where no human has ever been? Why would a spelunker care if someone found a new cave that has never been seen before? The same reason that mathematicians and hobbiest prime-hunters care about finding the largest known mersenne prime.
It has no immediate practical value, but it's always interesting when something new is discovered, especially when it's something that is the largest of its kind.
Gytis Barzdukas, director of product management in Microsoft's security business technology unit, said the free anti-spyware tools would only be available for a "personal edition" and made it clear that an enterprise version with management capabilities will carry a price tag.
"We're planning a refresh [of the anti-spyware beta] within the next few months. In the meantime, we're talking to enterprise customers, asking them about the management capabilities they need. Once we figure all that out, we will release a managed version," Barzdukas said in an interview with eWEEK.com.
Hashing functions are designed to explicitly prevent or make exceedingly difficult that sort of thing, so you can't just "work backwards" in that way. Maybe you'd like to review the algorithm...
NASA Statement on False Claim of Evidence of Life on Mars
News reports on February 16, 2005, that NASA scientists from Ames Research Center, Moffett Field, Calif., have found strong evidence that life may exist on Mars are incorrect.
NASA does not have any observational data from any current Mars missions that supports this claim. The work by the scientists mentioned in the reports cannot be used to directly infer anything about life on Mars, but may help formulate the strategy for how to search for martian life. Their research concerns extreme environments on Earth as analogs of possible environments on Mars. No research paper has been submitted by them to any scientific journal asserting martian life.
Domain names are ascii-only. What you are seeing is an example of IDN which includes a scheme to convert unicode into ascii so that you can have domain names with localized characters. The actual domain is http://xn--gba.com/ which is what (copyright symbol).com translates to in punycode.
I'm not trying to be disagreeable at all, and I don't really care about the timing of the posting of the article and the various replies. It doesn't matter to me if replies take days or weeks. I'm just trying to demonstrate how "cron+make" perhaps isn't the end-all solution to the question posed. A technical solution should stand on its own merits and I'm sad to say that "cron+make" works for a lot of cases but leaves the case of the original article poster in the dark.
How does that fit the requirements? Sure it can be hacked up to work, but consider that what he wants is not "run this at time X and follow these dependencies", it's "must run at any time after X and requires Y and Z to have happened." You can set a cron job to fire at X, but if Y and/or Z hasn't finished/happened yet what are you to do? Just sit there and wait? What if other things depend on this task? Will they all pile up into a heap of waiting processes? Can the waiting process really block on two independent events efficiently? What about 20, 50, 100? What if the dependent-task never happens, are you going to have the process kill itself eventually?
Or perhaps have the cron job poll every 5 minutes? Seems like that wouldn't scale if you have lots of processes. And because cron doesn't know about dependencies it still has to spawn something every 5 minutes even if the job has been run already -- even if it's just to check if the job has been run.
Both solutions seem wasteful.
I think what he was asking for is a scheduler that can handle both the time and the dependency checking from the same logic. Cron only knows to launch things at certain times and make only knows how to check dependencies. They are two seperate things that solve two seperate problems, in the great unix tradition, but it just seems to me very hackish to try to combine them when what you really want is a dispatcher/scheduler that knows about both. With make+cron because the two parts don't have knowledge of the other, you're basically back to polling every 'n' minutes to see if requirements have been met yet.
Mod parent up. The "not everyone knows anything" line is pure BS. I don't expect a beat cop to have memorized the statutes of intellectual property law, and I don't expect a family physician/general practictioner to know the treatments for obscure diseases. But I do expect that anyone that works for an IT department and maintains a publicly-facing website should know better than to use "username test, password test" to secure an interface that contains credit card and personal data. (Which is what happened in this case...)
I don't know why no one has posted it yet but here's a link to the text of the patent. The claimed date of filing seems to be 3-July-2000, when the provisional patent was filed.
Subject: Update: Outage of gcc.gnu.org / sourceware.org / sources.redhat.com Date: Sat, 5 Feb 2005 18:49:15 +0000 (GMT)
This is an update on the status of sourceware.org / gcc.gnu.org/ sources.redhat.com. It is being sent as "regular email" in the absence of working mailing lists.
As previously reported, the system suffered a hard drive failure. In the process of recovering from this, we discovered out-of-date RAID firmware which exacerbated the problems of recovery. This resulted in filesystem damage which made recovery of live data impossible and so we have been forced to restore completely from backup. Fortunately the backups are very recent. In particular, the CVS repository was backed up two hours before the system shutdown.
So, things are progressing but much of the backup is off-site, leading to some transfer delays. Once the backup is completed, the overseers and project maintainers will need to do some checking before the system can return to normal operation. The current rough estimate for bringing most things back online is tomorrow evening California time, or Monday morning UTC. There will probably not be another update mail before then.
Again, if there are any questions, e-mail Angela who is one of the sourceware overseers.
But that is about copy protection schemes, not about making copies. Basic copyright law determines how you can make copies, and the DMCA expands on that to make it a crime to circumvent measures meant to protect copyright. The above clarification relaxes the prohibition on circumventing copy-protection measures but it says nothing about basic copyright law, which is at the heart of the story author's premise.
Paul Vixie mantains a list of people offering inexpensive 1U colo services. You might start there and then try to search for testimonials or experiences with any that seems like good deals.
You should also check out the advertising forums on WebHostingTalk.com, where you will find many, many good deals on dedicated servers, colo, VPS, shared, etc.
The prices of the service listed in this article are, quite frankly, laughable. Their cheapest plan would be $43/month for a paultry 50GB of traffic. For around $50 to $75 you could get either a full 1Mbit (approx 316GB/month) 1U colo, or you could rent a dedicated server with probably a couple hundred GB metered. And in the latter you don't have to buy a machine.
For the $25 to $50 range get a VDS. You get all the control of a dedicated machine, without the cost. And you could most definitely do a heck of a lot better than a measly 50GB a month for a $40 or $50 VDS.
I just don't understand why you'd want to colo a mac mini. The small, sexy design is meaningless if it's locked in a datacenter, so you might as well just get a standard 1U server and be done with it. You have MANY more options for hosting a 1U or 2U server than some dinky little mac mini. Competition = better deals.
Before you advise people on how to use google you might want to learn how it works. Google has featured stemming for quite some time, and so you don't need to waste time with this '"moving company" OR "moving companies"' stuff. It's even mentioned in Google's basic help page.
Slashdot Mirror
...and so, in other words the switch had nothing to do with performance or stability, it was all about the presence of a much more developed and production-ready .NET toolchain and dev environment. That may be a valid argument for choosing win32 over linux, but it's not the reason the original poster thought it was ("oh it's so much more stable.")
Package versions are an unfair thing to pick on in that respect. The package version has to denote, among other things:
- The upstream baseline version of the source code
- Any 3rd party feature or bugfix patches applied
- Allow for version bumps when minor packaging issues arise (e.g. adding a "depends" or "requires")
While I agree that in some cases it does get out of hand, the vast majority of that is for packaging reasons where you need a unique version number to represent every package, even if the only thing that has changed is moving around the location of some man file or adding a README or something, or other minor tweaks that relate only to packaging issues and not actual upstream changes.
You may be able to write decent interfaces, or at least you think they are.
But the vast majority of flash-based sites that I have encountered make life harder. Several examples are:
- Tiny 6 point miniscule fixed fonts that I can't resize, contrary to what I've told my browser to use.
- Scrollbars that don't obey the mouse scroll whell and often times are not even real scrollbars (i.e. you HAVE to use the up/down buttons, you can't even drag the elevator - sometimes you can't even CLICK on the up/down buttons, you have to hover over them and WAIT for the text to scroll)
- Packing the text into a miniscule text area so that the presentation will work on granny's 640x480 display, meaning that to read more than a sentence or two I have to use some godawful nonstandard scrollbar, while the vast majority of my screen space is blank.
- Breaking the ability to easily bookmark a section of a page.
- Breaking the ability to use the back-button or open-link-in-new-tab features, which I use extensively.
If you call that better than what stock html gives me, then I'm afraid we disagree on what a user interface should be.
If you are one of the flash developers out there that have a clue and don't foister the above nonsense on your web visitors, then all the more power to you. But realize that this "hatred for flash" that you seem to have noticed is because the vast majority of people making flash DO violate one or all of the above annoyances. If that's the case then your beef is with these incompetants soiling the reputation of flash, and not for the rest of us that have been burned by it too many times and have dismissed it.
Check out http://backports.org/.
Microsoft used the BSD networking code in the kernel for its first version of Windows NT. Starting with NT v3.5 in 1994, the TCP/IP stack was completely rewritten from scratch by Microsoft using no BSD code. There is still some BSD code in use in the userspace utilities, like ping, telnet, et cetera.
Can we PLEASE put to rest this notion that any version of Windows made in the last 10 years has any kind of BSD TCP/IP stack code in it?
SOURCE
Umm, on most compilers "int" is a 32 bit value. I don't know what it is on yours. But a 32-bit integer can not come anywhere close to representing 1e12, so essentially what you were asking for was an infinite loop, or whatever loop results from casting (float)1e12 into an integer, which is probably undefined and thus the compiler can't optimise there.
You think the spyware EULAs are bad, try this one for a popular torrent site: 18,554 words or 33 printed pages at 12 pt font and standard margins.
To be fair it's not their EULA, but the one that the original torrentbits web site came up with. Since TB shared its php site source code there are a lot of torrent sites that use a similar or somewhat modified site backend, including the monster EULA.
I think they did it rather as a joke, to make sure that any enforcement agencies reading it would at least have to pay for a couple of extra lawyer-hours of time if they hired councel.
I prefer WinRAR's interface to Winzip, so I install WinRAR and have it handle all compressed archive types (.zip, .rar, .arc, etc.) I'm sure I am not alone in this habit, as WinRAR is extremely popular. There are other programs as well (7-zip, Power Archiver) that handle all the common compression formats and are usually installed in a "use for everything" manner.
/m avoids temptation to take a dig at the crapulence and outdated-ness of winzip.
My point is that even if you have never encoutered a RAR file previously, there's a chance that your current archiver supports it. Or alternatively put, "lots of people don't use Winzip."
And car dealerships always love cash because that means all the money from the sale is going to them and not partly to a bank over a period of 5 years (or at least that's my theory, although I do know for a fact that they love cash.)
They may like motivated buyers (i.e. those with cash in hand) but they most certainly don't like missing out on arranging the financing. They have deals with the banks with which they're affiliated, so when you walk in there and arrange financing with them they get a cut of the deal. They make a very large portion of their total profits off these arrangements with lenders and banks, so they most definitely get shafted if you pay cash (or get a loan from your own bank/credit union before walking into the dealership for that matter.)
Why would a mountaineer care if someone sets a record for highest mountain climbed? Why would an explorer care if someone pioneered a part of the antartic where no human has ever been? Why would a spelunker care if someone found a new cave that has never been seen before? The same reason that mathematicians and hobbiest prime-hunters care about finding the largest known mersenne prime.
It has no immediate practical value, but it's always interesting when something new is discovered, especially when it's something that is the largest of its kind.
Source: http://www.eweek.com/article2/0,1759,1765389,00.a
I'm no crypto expert...
You could have stopped right there, honestly...
Hashing functions are designed to explicitly prevent or make exceedingly difficult that sort of thing, so you can't just "work backwards" in that way. Maybe you'd like to review the algorithm...
Source: http://www.nasa.gov/home/hqnews/2005/feb/HQ_05052
That is a great solution, but it's the antithesis of "turnkey" and is likely the complete opposite of what the "ask slashdot" poster was looking for.
That's not a link, that's a URL, and it's perverted by slashcode which adds spaces to words that are too long, making it useless for copy and paste.
6 073
A link would be something like http://www.mozillazine.org/talkback.html?article=
Domain names are ascii-only. What you are seeing is an example of IDN which includes a scheme to convert unicode into ascii so that you can have domain names with localized characters. The actual domain is http://xn--gba.com/ which is what (copyright symbol).com translates to in punycode.
I'm not trying to be disagreeable at all, and I don't really care about the timing of the posting of the article and the various replies. It doesn't matter to me if replies take days or weeks. I'm just trying to demonstrate how "cron+make" perhaps isn't the end-all solution to the question posed. A technical solution should stand on its own merits and I'm sad to say that "cron+make" works for a lot of cases but leaves the case of the original article poster in the dark.
How does that fit the requirements? Sure it can be hacked up to work, but consider that what he wants is not "run this at time X and follow these dependencies", it's "must run at any time after X and requires Y and Z to have happened." You can set a cron job to fire at X, but if Y and/or Z hasn't finished/happened yet what are you to do? Just sit there and wait? What if other things depend on this task? Will they all pile up into a heap of waiting processes? Can the waiting process really block on two independent events efficiently? What about 20, 50, 100? What if the dependent-task never happens, are you going to have the process kill itself eventually?
Or perhaps have the cron job poll every 5 minutes? Seems like that wouldn't scale if you have lots of processes. And because cron doesn't know about dependencies it still has to spawn something every 5 minutes even if the job has been run already -- even if it's just to check if the job has been run.
Both solutions seem wasteful.
I think what he was asking for is a scheduler that can handle both the time and the dependency checking from the same logic. Cron only knows to launch things at certain times and make only knows how to check dependencies. They are two seperate things that solve two seperate problems, in the great unix tradition, but it just seems to me very hackish to try to combine them when what you really want is a dispatcher/scheduler that knows about both. With make+cron because the two parts don't have knowledge of the other, you're basically back to polling every 'n' minutes to see if requirements have been met yet.
It's the name of the release group that did the capture. You might want to familiarize yourself with NFOs if this makes no sense.
Mod parent up. The "not everyone knows anything" line is pure BS. I don't expect a beat cop to have memorized the statutes of intellectual property law, and I don't expect a family physician/general practictioner to know the treatments for obscure diseases. But I do expect that anyone that works for an IT department and maintains a publicly-facing website should know better than to use "username test, password test" to secure an interface that contains credit card and personal data. (Which is what happened in this case...)
I don't know why no one has posted it yet but here's a link to the text of the patent. The claimed date of filing seems to be 3-July-2000, when the provisional patent was filed.
But that is about copy protection schemes, not about making copies. Basic copyright law determines how you can make copies, and the DMCA expands on that to make it a crime to circumvent measures meant to protect copyright. The above clarification relaxes the prohibition on circumventing copy-protection measures but it says nothing about basic copyright law, which is at the heart of the story author's premise.
Paul Vixie mantains a list of people offering inexpensive 1U colo services. You might start there and then try to search for testimonials or experiences with any that seems like good deals.
You should also check out the advertising forums on WebHostingTalk.com, where you will find many, many good deals on dedicated servers, colo, VPS, shared, etc.
The prices of the service listed in this article are, quite frankly, laughable. Their cheapest plan would be $43/month for a paultry 50GB of traffic. For around $50 to $75 you could get either a full 1Mbit (approx 316GB/month) 1U colo, or you could rent a dedicated server with probably a couple hundred GB metered. And in the latter you don't have to buy a machine.
For the $25 to $50 range get a VDS. You get all the control of a dedicated machine, without the cost. And you could most definitely do a heck of a lot better than a measly 50GB a month for a $40 or $50 VDS.
I just don't understand why you'd want to colo a mac mini. The small, sexy design is meaningless if it's locked in a datacenter, so you might as well just get a standard 1U server and be done with it. You have MANY more options for hosting a 1U or 2U server than some dinky little mac mini. Competition = better deals.
Before you advise people on how to use google you might want to learn how it works. Google has featured stemming for quite some time, and so you don't need to waste time with this '"moving company" OR "moving companies"' stuff. It's even mentioned in Google's basic help page.