OK so I click on the link and what do I see... no useful content whatsoever, just some marketspeak about an 'exclusive social experience'. Why would I bother spending five minutes to sign up? Wikipedia shows how to do it right: the front page is interesting reading, regularly updated, full of featured articles and 'did you know' to draw people into the site.
Stephen Hawking isn't very aesthetically pleasing but he's rarely wrong.
I thought one of the main tests scientists use when evaluating a new theory is whether it 'looks right' or its 'beauty' (which could be another way of saying simplicity). And in mathematics, conjectures are often judged by how pretty the equation seems - to decide which ones are likely to be right, and worth trying to make a proof for. Certainly in programming the aesthetic quality of a program is the most important thing after making sure it passes the tests.
Your analogy of insects is interesting: it does appear that bugs which are ugly or make us instinctively go 'yuck' are also those we'd want to avoid because they are parasites or spread diseases. A picture of a tick or flea evokes feelings of ugliness at some instinctual level.
The story takes the angle of the little guy bravely battling against unfeeling big government. And that's surely how pirate operators and their listeners see it. But if the range of your FM transmitter is only a few miles, why is it a federal matter? These frequency bands should be left to the states to allocate as they wish.
RCS can be quite handy to use for configuration files, so you can have/etc/fstab and/etc/fstab,v. It's also used as a backend for other programs, like CVS (in the early days - CVS still uses RCS format files in its repository but has its own code to handle them) or to provide version history in TWiki.
Another approach to try and prevent this might be to get the browser not to send conditional GET requests *at all* and to just reload silently from cache.
Back when I used a modem I had the wwwoffle proxy server set to always used cached pages whenever possible - the only way to get an updated version from the site was to hit Reload. It was nice and fast, and sometimes useful to be able to still browse a site that had disappeared in the real world, although on hitting Reload your precious page would disappear.
Is there a correlation between which browser people use and how worth reading their comments are? To measure it, you could use moderation score as a very rough approximation to 'worth reading'.
I still insist on a local email client that stores all of my email/on/my/computer/.
In what format? If it's stored as a big blob of binary garbage, you're just as much held to ransom as if it were on a remote server. You still can't get at your data except by going through the 'official channel', in this case running that particular mail program and hoping it doesn't crash or corrupt its data store.
Insist on mail stored in a readable format like mbox files or maildirs!
I think most of those environmental costs are already factored into what you pay, after all, gold costs money, companies pay to have their industrial waste disposed of safely, electricity isn't free, and you must pay to buy land and building materials. There are surely externalities that aren't accounted for, but they're pretty small in relation to the things that already require large wads of cash.
Sure, average users don't care about 64-bit performance. But average users don't care about performance *at all*. If you just want to use online poker, MSN and websites, the speed of your CPU is nowadays completely irrelevant. Anything you buy will be ridiculously overspecified for those tasks.
Gaming is about the only mainstream Joe Consumer application that cares about hardware speed, and even that is mostly dominated by the video card not the CPU.
So if you're going to bother making benchmarks, you might as well cater to the people who care about performance, which means those doing number-crunching (who want 64 bit because it's faster) or just possibly people doing endless kernel compiles and ssh tunnelling (who will also be on 64 bit because they can).
Finding and fixing an individual vulnerability is necessary, but not that interesting. What's more important is to look at what kind of bug it is, see if it occurs elsewhere, and see what might be done to stop it happening in future. Often that means providing programmers with a safer mechanism that makes it harder to introduce bugs (or makes bugs that do creep in be harder to exploit). The OpenBSD project has justly become famous for this kind of auditing, for example, they found in auditing that copying of C-style strings often led to exploitable buffer overruns, and developed strlcpy() as a safer alternative.
So what does the Firefox list tell us?
A large number of defects resulted from the code not checking for null after memory was allocated.
This could be fixed by using a memory allocation function that throws an exception on out-of-memory, rather than relying on the programmer to remember to check every call. If the programmer forgets to use a try/catch block to check for exceptions, the worst that happens is that the exception propagates upwards to a higher-level 'catch-all' routine.
In addition, there were many cases where the return value of functions designed to return null were not checked prior to dereferencing.
Again, this shows that returning a special value such as null is not always the best thing to do. If it's likely that programmers will forget to check it (perhaps because null is only returned in error conditions, and doesn't occur during ordinary use), then an exception would be a better way to signal an error. It also lets you give more information about exactly what the error was; you can have different exceptions carrying different messages for different things that went wrong, but null is just null.
Hmm, but later on in TFA I see a comment from a Mozilla developer:
With most of these tools the signal:noise ratio is very high. For example, most of these "dereferencing null" cases are either handled automatically by C++ template wrappers that do smart pointer management. Many of these "potential" memory leaks are handled automatically by XPCOM's refcounting.
So it looks like the Mozilla folk have already thought about these problems and done the Right Thing.
Yup - unfortunately some MTAs can't do this and can (by design) only send soft bounces, which makes them unsuitable for today's spam-ridden world. It would be nice if MTA comparison reviews mentioned this.
Suppose your normal mail is 90% spam and 10% legit (a pretty reasonable figure if your address is published on the web; my mail is two orders of magnitude worse than that). Then suppose that your spam filter is 90% accurate, in that it tags 90% of spam as spam, and only tags 10% of legitimate mail as spam. (Again, real spam filters usually have much lower false positive and false negative rates than this, after a bit of training.) Then for every one genuine 'rejected' message you send out, you'll send 81 completely bogus messages to forged 'From' addresses. Indeed, your server will be producing almost as much spam as it receives. And once people start getting spam from your mail server, they'll tag it as spam in their mail programs, and your domain and the keywords you use in your rejected messages will get associated with spamminess. You might even get onto a few blacklists. Legitimate mail from your domain will start to be filtered as spam by other people, and as for that one genuine rejection message you were so keen to send out... it's likely to be silently dropped into the recipient's spam folder.
Now I am doom-mongering a bit here. I do still get 'your mail was filtered out' messages from various people, and a lot of the time they don't get marked as spam because each one is different. I don't see any genuine ones, BTW, just bogus ones from worms spoofing mail from my address. The point stands that if you try to send a rejection message for every possible-spam you receive, almost all of them will be bogus and you will be pumping out almost as much sewage as you receive.
Agreed. I'm considering filing bug reports against free software that does this; however, it'd just be my opinion against the author's, and it would be a little bit obnoxious of me. An explanatory web page on FSF's site, saying why click-through agreements are counterproductive for free software, would make it easier to get these fixed.
You can go to prison for contempt of court, but I think it's very rare for it to happen to journalists. Hmm, this Telegraph leading article has more on the subject.
It's not the government that threatens people here. It is individual judges presiding over a criminal case. If you publish a story concerning the defendant or details of the case, the judge can get upset that it might influence the jury and so journalists can be prosecuted for contempt of court. I don't think journalists go to jail, they just get fined, but still British newspapers are careful not to discuss details of cases that are sub judice.
Judges do not throw around the contempt of court charges just for fun; I think they only do it if they feel there is a serious risk that a fair trial might be prejudiced by the press coverage. If that happens then generally the criminal trial in progress is abandoned and the defendant is automatically found not guilty.
All this may be completely unnecessary, after all the US and perhaps other countries have jury trials without worrying that press coverage might influence the jury.
I've never been stupid enough to buy a DRM encumbered piece of content.
There is no such thing as DRM-encumbered content; only broken applications that do not work fully with certain files. That said you may want to protest by not buying from publishers who try to insist that only DRM-broken programs can view their media.
Slashdot Mirror
You can start programming in Perl 6 today using Pugs.
Yup, the GPL was carved onto stone tablets.
OK so I click on the link and what do I see... no useful content whatsoever, just some marketspeak about an 'exclusive social experience'. Why would I bother spending five minutes to sign up? Wikipedia shows how to do it right: the front page is interesting reading, regularly updated, full of featured articles and 'did you know' to draw people into the site.
Your analogy of insects is interesting: it does appear that bugs which are ugly or make us instinctively go 'yuck' are also those we'd want to avoid because they are parasites or spread diseases. A picture of a tick or flea evokes feelings of ugliness at some instinctual level.
The story takes the angle of the little guy bravely battling against unfeeling big government. And that's surely how pirate operators and their listeners see it. But if the range of your FM transmitter is only a few miles, why is it a federal matter? These frequency bands should be left to the states to allocate as they wish.
RCS can be quite handy to use for configuration files, so you can have /etc/fstab and /etc/fstab,v. It's also used as a backend for other programs, like CVS (in the early days - CVS still uses RCS format files in its repository but has its own code to handle them) or to provide version history in TWiki.
'in your opinion'... how many times have you tried it?
Er, wait - nobody answer that...
Is there a correlation between which browser people use and how worth reading their comments are? To measure it, you could use moderation score as a very rough approximation to 'worth reading'.
The book was previously reviewed on Slashdot. Better just to note what's changed in the new edition.
In what format? If it's stored as a big blob of binary garbage, you're just as much held to ransom as if it were on a remote server. You still can't get at your data except by going through the 'official channel', in this case running that particular mail program and hoping it doesn't crash or corrupt its data store.
Insist on mail stored in a readable format like mbox files or maildirs!
I think most of those environmental costs are already factored into what you pay, after all, gold costs money, companies pay to have their industrial waste disposed of safely, electricity isn't free, and you must pay to buy land and building materials. There are surely externalities that aren't accounted for, but they're pretty small in relation to the things that already require large wads of cash.
Sure, average users don't care about 64-bit performance. But average users don't care about performance *at all*. If you just want to use online poker, MSN and websites, the speed of your CPU is nowadays completely irrelevant. Anything you buy will be ridiculously overspecified for those tasks.
Gaming is about the only mainstream Joe Consumer application that cares about hardware speed, and even that is mostly dominated by the video card not the CPU.
So if you're going to bother making benchmarks, you might as well cater to the people who care about performance, which means those doing number-crunching (who want 64 bit because it's faster) or just possibly people doing endless kernel compiles and ssh tunnelling (who will also be on 64 bit because they can).
So what does the Firefox list tell us?
This could be fixed by using a memory allocation function that throws an exception on out-of-memory, rather than relying on the programmer to remember to check every call. If the programmer forgets to use a try/catch block to check for exceptions, the worst that happens is that the exception propagates upwards to a higher-level 'catch-all' routine.
Again, this shows that returning a special value such as null is not always the best thing to do. If it's likely that programmers will forget to check it (perhaps because null is only returned in error conditions, and doesn't occur during ordinary use), then an exception would be a better way to signal an error. It also lets you give more information about exactly what the error was; you can have different exceptions carrying different messages for different things that went wrong, but null is just null.
Hmm, but later on in TFA I see a comment from a Mozilla developer:
So it looks like the Mozilla folk have already thought about these problems and done the Right Thing.
Yup - unfortunately some MTAs can't do this and can (by design) only send soft bounces, which makes them unsuitable for today's spam-ridden world. It would be nice if MTA comparison reviews mentioned this.
Suppose your normal mail is 90% spam and 10% legit (a pretty reasonable figure if your address is published on the web; my mail is two orders of magnitude worse than that). Then suppose that your spam filter is 90% accurate, in that it tags 90% of spam as spam, and only tags 10% of legitimate mail as spam. (Again, real spam filters usually have much lower false positive and false negative rates than this, after a bit of training.) Then for every one genuine 'rejected' message you send out, you'll send 81 completely bogus messages to forged 'From' addresses. Indeed, your server will be producing almost as much spam as it receives. And once people start getting spam from your mail server, they'll tag it as spam in their mail programs, and your domain and the keywords you use in your rejected messages will get associated with spamminess. You might even get onto a few blacklists. Legitimate mail from your domain will start to be filtered as spam by other people, and as for that one genuine rejection message you were so keen to send out... it's likely to be silently dropped into the recipient's spam folder.
Now I am doom-mongering a bit here. I do still get 'your mail was filtered out' messages from various people, and a lot of the time they don't get marked as spam because each one is different. I don't see any genuine ones, BTW, just bogus ones from worms spoofing mail from my address. The point stands that if you try to send a rejection message for every possible-spam you receive, almost all of them will be bogus and you will be pumping out almost as much sewage as you receive.
Antivirus companies used to be particularly bad at this; see Anti-Virus Companies: Tenacious Spammers.
If you think the software isn't entirely free, contact Sourceforge. Their conditions require that all hosted projects be free software.
Agreed. I'm considering filing bug reports against free software that does this; however, it'd just be my opinion against the author's, and it would be a little bit obnoxious of me. An explanatory web page on FSF's site, saying why click-through agreements are counterproductive for free software, would make it easier to get these fixed.
You can go to prison for contempt of court, but I think it's very rare for it to happen to journalists. Hmm, this Telegraph leading article has more on the subject.
It's not the government that threatens people here. It is individual judges presiding over a criminal case. If you publish a story concerning the defendant or details of the case, the judge can get upset that it might influence the jury and so journalists can be prosecuted for contempt of court. I don't think journalists go to jail, they just get fined, but still British newspapers are careful not to discuss details of cases that are sub judice.
Judges do not throw around the contempt of court charges just for fun; I think they only do it if they feel there is a serious risk that a fair trial might be prejudiced by the press coverage. If that happens then generally the criminal trial in progress is abandoned and the defendant is automatically found not guilty.
All this may be completely unnecessary, after all the US and perhaps other countries have jury trials without worrying that press coverage might influence the jury.
There is no such thing as DRM-encumbered content; only broken applications that do not work fully with certain files. That said you may want to protest by not buying from publishers who try to insist that only DRM-broken programs can view their media.
IBM calls it a hard drive? Nooo....
You always used to be able to rely on IBM for consistently curmudgeonly naming, like calling a motherboard a planar or a floppy disk a diskette.