It should be more than just reset. For example, connect power while the reset button is held down and hold it for 3 seconds. A mere reset should not erase the configuration. And even then, the default start menu should offer to restore the original configuration if there is one present with a valid checksum.
That's pretty much what the angif library does. It just doesn't include the special animation block that Netscape defined as an extension to GIF, since it isn't doing animation... unless you specifically want to make an animated true color GIF. Also, angif recursively slices the image until it gets to a level where there are not more than 256 colors (which stops well before it gets down to a single pixel). This explains the funny loading pattern you get in true color mode (the order of the blocks).
So who recorded the entire event at the best resolution (spatially and temporally) and is making those images available free in an open format (preferrably discrete PNG image files, but JPEG, GIF (monochrome only unless you know how to do true color in GIF like I do), or MPEG would be OK) online?
Whoever does have access to the source code is who should be responsible for fixing it according to the requirements of whatever manufacturers are incorporating it into their products. If Netgear's own staff didn't write it, they either know who did, or know who knows, and can track it back to there and demand a secure version.
Very likely, someone wanted this feature for support reasons (let's not jump on the spy reason bandwagon just yet). Someone, somewhere, might be whining that support costs will go up, and customer satisfaction will go down, because more units will have to be returned to be fixed, rather than guide the customer through the setup (which is not going to work on the 75% of the population that can't follow directions).
And that document specifically declines (see the FAQ at the bottom) to discuss the legal liabilities Netgear incurred. I think there is substantial interest in that, even if it is simply a statement like "Netgear is covering all costs incurred". Taxpayers of Wisconsin would probably want to know. Students who pay tuition would probably want to know.
So now my web browser will get fatter as all these new bells and whistles I don't need will be bundled in. What we need is to separate all these features into their own application, and have simply a small framework (a version of which could replace X and go directly to the video) that manages the screen. At least this way I can kill those particular processes (like Flash) that usually need to be killed.
Come on, seriously, putting all these application capabilities in a web browser isn't conceptually much different than a windowing system (besides the specific API and protocol differences). Pretty soon we'll do everything in a (so called) web browser super app and the windowing system will do little more than just start this one beast (and thus be a relatively lame layer). Why not merge these things and make a complete video driver, window system, and apps manager in a uniform design?
Too much liberty in the implementation and a lack of a properly defined standard is the main problem.
Standards should never dictate human user interfaces. Instead, human users should have a choice of the interface they prefer. You should never impose your preference on me, and I should never impose my preference on you. Standards should specify how the applications access the cut and paste buffer area, only.
It is my understanding that it can only work one way in Microsoft Windows, but that X (that's the name of it, just "X") provides the mechanisms to exchange buffer data and imposes no particular means to interface the human user. The window managers do that (usually), and there are plenty of them to choose from. If you want a Ctrl-C/Ctrl-V style cut/paste interface, there might be a window manager already written to do it (and probably lots of other Microsoftisms at the same time). I wouldn't know because I've had no need to look for one. But if there is not one, you're certainly welcome to grab an existing window manager implementation (most of them are open source, free (as in beer), free (as in speech), and free (as a bird), allowing you to hack the source to make it behave as you prefer, and share your accomplishments with your friends and enemies).
SPFy isn't the total solution, but it certainly can help cut some spam out of my mailbox (and those of my users). I do think people should be made well aware that this isn't going to ultimately solve the problem. Instead, some providers need to have the "Email Death Penalty" applied to get them to carry out the proper enforcement against abuses from their customers. One problem is that many of the providers that host spammers are used by the "collateral damage" whiners seen in/. and elsewhere. But at some people we do need to apply the EDP to say the top 20 spam hosters and tell the whiners to go to {hell, another ISP}[their choice].
And SPF also increases the server resource usage a small amount by going through the motions. So SPF is a "keep spam outa my box" solution, but it is not a "stop the spamming abuse" solution that (IMHO) too few people are making any effort at. Somehow, we need to get ISPs to stop letting spammers have access, but so far the only means to accomplish that being put forward results in lots of people whining about the collateral damage (instead of changing to another ISP like they should).
For cutting the bandwidth usage, you really need to implement 2 features on a site-wide basis. 1: Blacklists. *Harsh* blacklists, that bounce email from sites that not merely send spam but that allow spam or refuse to fix open relays.
What about the providers that continue to keep the spammers connected? Is it too *harsh* to also block them (and thus make collateral damage of their other customers)?
I understand SPF will allow me to refuse email from sites that are not designated (if the sender domain publishes appropriate SPF records), and allows me to publish records to allow those using it to refuse attempts to forge senders in my domains. I already do publish SPF data, and will soon be adding SPF checking in my mail servers. But this only stops spam (and backscatter) from getting into my mailbox. What can be done to keep spammers from even trying to connect to any of my servers?
At a minimum they can spy on your traffic. Is all your traffic encrypted? Even the encrypted traffic tells them where you are going.
Now, if you were to run a NetBSD server in Russia, and always access everything by going through it via an encrypted tunnel on your home NetBSD server, it might be harder for them to track you for a couple days.
Crashing is an unpleasant thing. But so is thrashing. Of course you need to have enough RAM to handle the things you are doing. Swap can then be used for "peak shaving" for those not-so-often times when you exceed expected needs. One problem is that when processes do a lot of disk writing, the Linux kernel tries to write things to RAM first, then later to disk... preferring to write "idle" process memory to disk (swap) first instead. Linux needs to have a limit on the amount of RAM used for write caching, and a reserve of RAM for process memory only. I want to be able to keep the normal amount of processes in RAM regardless of how many hundred of gigabytes a few of them are writing to disk files.
So fix the cause, not the symptom ...
on
Is Swap Necessary?
·
· Score: 1
So fix the cause, not the symptom... don't use bloated programs.
The amount of swap space to configure should be based not on the size of your RAM, but the speed of your hard drive and transfer bus. Choose a size that can be swapped (in or out) in less than 10 seconds (or whatever your level of tolerance is).
So why try to fix the symptom (heavy swapping). Instead, fix the cause (bloat). Let's go back to efficient programming that only 5% of C programmers know how to do (using lean code and scalable algorithms).
Steve Jobs said that regarding the Lisa or whatever it was that came out before the Mac. He said that about 256k, I think when someone raised the issue that the new IBM PC could be expanded to 640k.
Use port 587 instead. That's the message submission protocol. It's a subset of SMTP-AUTH+TLS over a different port designated for this very purpose. Outlook and other mail agents support it (if you can enable TLS and specify a port number, you can use it). Any 3rd party mail provider destined to stay in business has this available.
First of all, let me say this. A router (any packet routing component of the infrastructure) that uses sequential testing of access lists is a piece of crap. But then you probably already know this. And yes, this probably describes 99% of the equipment still out there. Ideally, a mechanism to lookup access lists should be done just the same way as routing tables are looked up (which for some high end stuff is done in hardware, too).
There are two basic categories of people affected by blocking port 25:
1. Business class customers with mail servers who do not wish to smart host forward through the ISP mail servers. The solution for these is simple: you have a separate pool of IP address space for them. This would allow selectively blocking or unblocking port 25 based on the pool class, rather than individual customer.
2. People wanting to send mail through a 3rd party mail provider (mail gets through better if the mail server sending their mail to the MX host has appropriate designation information). But port 25 is not the proper way to use this kind of service. Instead, the message submission protocol (which is essentially a subset of SMTP-AUTH/TLS on its own port 587) should be used. Outlook supports it. Most mail server software supports it (especially with add-ons like SASL). Your tech support people should be trained to recognize these users and instruct them to use the message submission protocol with their 3rd party mail provider.
Slashdot Mirror
... except when it's connecting to the SETI server to exchange another work unit.
It should be more than just reset. For example, connect power while the reset button is held down and hold it for 3 seconds. A mere reset should not erase the configuration. And even then, the default start menu should offer to restore the original configuration if there is one present with a valid checksum.
That's pretty much what the angif library does. It just doesn't include the special animation block that Netscape defined as an extension to GIF, since it isn't doing animation ... unless you specifically want to make an animated true color GIF. Also, angif recursively slices the image until it gets to a level where there are not more than 256 colors (which stops well before it gets down to a single pixel). This explains the funny loading pattern you get in true color mode (the order of the blocks).
Actually, I wrote it.
So who recorded the entire event at the best resolution (spatially and temporally) and is making those images available free in an open format (preferrably discrete PNG image files, but JPEG, GIF (monochrome only unless you know how to do true color in GIF like I do), or MPEG would be OK) online?
Whoever does have access to the source code is who should be responsible for fixing it according to the requirements of whatever manufacturers are incorporating it into their products. If Netgear's own staff didn't write it, they either know who did, or know who knows, and can track it back to there and demand a secure version.
Very likely, someone wanted this feature for support reasons (let's not jump on the spy reason bandwagon just yet). Someone, somewhere, might be whining that support costs will go up, and customer satisfaction will go down, because more units will have to be returned to be fixed, rather than guide the customer through the setup (which is not going to work on the 75% of the population that can't follow directions).
And that document specifically declines (see the FAQ at the bottom) to discuss the legal liabilities Netgear incurred. I think there is substantial interest in that, even if it is simply a statement like "Netgear is covering all costs incurred". Taxpayers of Wisconsin would probably want to know. Students who pay tuition would probably want to know.
So now my web browser will get fatter as all these new bells and whistles I don't need will be bundled in. What we need is to separate all these features into their own application, and have simply a small framework (a version of which could replace X and go directly to the video) that manages the screen. At least this way I can kill those particular processes (like Flash) that usually need to be killed.
Come on, seriously, putting all these application capabilities in a web browser isn't conceptually much different than a windowing system (besides the specific API and protocol differences). Pretty soon we'll do everything in a (so called) web browser super app and the windowing system will do little more than just start this one beast (and thus be a relatively lame layer). Why not merge these things and make a complete video driver, window system, and apps manager in a uniform design?
Standards should never dictate human user interfaces. Instead, human users should have a choice of the interface they prefer. You should never impose your preference on me, and I should never impose my preference on you. Standards should specify how the applications access the cut and paste buffer area, only.
It is my understanding that it can only work one way in Microsoft Windows, but that X (that's the name of it, just "X") provides the mechanisms to exchange buffer data and imposes no particular means to interface the human user. The window managers do that (usually), and there are plenty of them to choose from. If you want a Ctrl-C/Ctrl-V style cut/paste interface, there might be a window manager already written to do it (and probably lots of other Microsoftisms at the same time). I wouldn't know because I've had no need to look for one. But if there is not one, you're certainly welcome to grab an existing window manager implementation (most of them are open source, free (as in beer), free (as in speech), and free (as a bird), allowing you to hack the source to make it behave as you prefer, and share your accomplishments with your friends and enemies).
A floppy drive is recommended for CWRU students? Aren't those obsolete? I haven't used floppies since last century.
SPFy isn't the total solution, but it certainly can help cut some spam out of my mailbox (and those of my users). I do think people should be made well aware that this isn't going to ultimately solve the problem. Instead, some providers need to have the "Email Death Penalty" applied to get them to carry out the proper enforcement against abuses from their customers. One problem is that many of the providers that host spammers are used by the "collateral damage" whiners seen in /. and elsewhere. But at some people we do need to apply the EDP to say the top 20 spam hosters and tell the whiners to go to {hell, another ISP}[their choice].
And SPF also increases the server resource usage a small amount by going through the motions. So SPF is a "keep spam outa my box" solution, but it is not a "stop the spamming abuse" solution that (IMHO) too few people are making any effort at. Somehow, we need to get ISPs to stop letting spammers have access, but so far the only means to accomplish that being put forward results in lots of people whining about the collateral damage (instead of changing to another ISP like they should).
What about the providers that continue to keep the spammers connected? Is it too *harsh* to also block them (and thus make collateral damage of their other customers)?
I understand SPF will allow me to refuse email from sites that are not designated (if the sender domain publishes appropriate SPF records), and allows me to publish records to allow those using it to refuse attempts to forge senders in my domains. I already do publish SPF data, and will soon be adding SPF checking in my mail servers. But this only stops spam (and backscatter) from getting into my mailbox. What can be done to keep spammers from even trying to connect to any of my servers?
And how does blackholing the spammer IPs get them to stop trying?
So how does this help me reduce the amount of bandwidth and server resources used by spammers who continue to try sending spam to me and my users?
At a minimum they can spy on your traffic. Is all your traffic encrypted? Even the encrypted traffic tells them where you are going.
Now, if you were to run a NetBSD server in Russia, and always access everything by going through it via an encrypted tunnel on your home NetBSD server, it might be harder for them to track you for a couple days.
Conspiracy theorists are needed. Who else is willing to take random jabs in the hopes of finding something they had no real reason to know was there?
Choose two.
Crashing is an unpleasant thing. But so is thrashing. Of course you need to have enough RAM to handle the things you are doing. Swap can then be used for "peak shaving" for those not-so-often times when you exceed expected needs. One problem is that when processes do a lot of disk writing, the Linux kernel tries to write things to RAM first, then later to disk ... preferring to write "idle" process memory to disk (swap) first instead. Linux needs to have a limit on the amount of RAM used for write caching, and a reserve of RAM for process memory only. I want to be able to keep the normal amount of processes in RAM regardless of how many hundred of gigabytes a few of them are writing to disk files.
So fix the cause, not the symptom ... don't use bloated programs.
The amount of swap space to configure should be based not on the size of your RAM, but the speed of your hard drive and transfer bus. Choose a size that can be swapped (in or out) in less than 10 seconds (or whatever your level of tolerance is).
So why try to fix the symptom (heavy swapping). Instead, fix the cause (bloat). Let's go back to efficient programming that only 5% of C programmers know how to do (using lean code and scalable algorithms).
Steve Jobs said that regarding the Lisa or whatever it was that came out before the Mac. He said that about 256k, I think when someone raised the issue that the new IBM PC could be expanded to 640k.
Use port 587 instead. That's the message submission protocol. It's a subset of SMTP-AUTH+TLS over a different port designated for this very purpose. Outlook and other mail agents support it (if you can enable TLS and specify a port number, you can use it). Any 3rd party mail provider destined to stay in business has this available.
First of all, let me say this. A router (any packet routing component of the infrastructure) that uses sequential testing of access lists is a piece of crap. But then you probably already know this. And yes, this probably describes 99% of the equipment still out there. Ideally, a mechanism to lookup access lists should be done just the same way as routing tables are looked up (which for some high end stuff is done in hardware, too).
There are two basic categories of people affected by blocking port 25:
1. Business class customers with mail servers who do not wish to smart host forward through the ISP mail servers. The solution for these is simple: you have a separate pool of IP address space for them. This would allow selectively blocking or unblocking port 25 based on the pool class, rather than individual customer.
2. People wanting to send mail through a 3rd party mail provider (mail gets through better if the mail server sending their mail to the MX host has appropriate designation information). But port 25 is not the proper way to use this kind of service. Instead, the message submission protocol (which is essentially a subset of SMTP-AUTH/TLS on its own port 587) should be used. Outlook supports it. Most mail server software supports it (especially with add-ons like SASL). Your tech support people should be trained to recognize these users and instruct them to use the message submission protocol with their 3rd party mail provider.