Programs compiled with gcc don't have to be GPLed. If they did, Apple wouldn't be using it in Mac OS X (the Darwin kernel is open-source but not GPL-compatible, the high-level bits like Aqua are proprietary.)
The community around Polycount tends to be strictly opposed to plagiarism, and it seems to a good idea to ask authors' permission before doing anything to their model (as well as being courteous - they like to know that their models are appreciated:-)
Some models on Polycount don't actually have a permissions statement, meaning that if you want to be picky, it may not even be legal for Polycount to distribute them. Some have non-commercial clauses in the readme, or other encumberances.
Some Q3 models use, or are based on, Quake 3 stock animations - I'm not sure about the copyright status of these, and whether they're a large enough proportion that the model counts as a derived work.
As for "Free" rather than "free", I haven't seen a model on Polycount yet that's freely modifiable, or meets the FSF or OSI guidelines for free software/open source software - so if you want open-source code combined with unmodifiable graphics, that's fine, but if you want your artwork to be licensed in the same way as your code, Polycount is unlikely to help you unless you specifically ask the authors for more permission.
Forcing one particular format is a recipe for disaster - as soon as a feature it doesn't support comes along, you need a new law.
What is worth forcing is that the format is documented sufficiently well that it can be reimplemented.
In this particular case, the format is easy to document, because you can just say "It's XML, it conforms to specifications A, B and C, and the whole thing is in a zip file as described by RFC 1234" - so all you actually need to document is the particular "application" (sub-language) of XML that you're using, preferably along with forwards-compatibility notes like "unknown elements are a fatal error" or "unknown attributes must be ignored" or whatever you like.
If you have a law forcing data to be in properly-documented formats, proprietary software companies are free to implement apps that work with that format, without open-sourcing anything.
They'll just have to compete (with open-source apps, and with other proprietary apps) on features, ease of use, reliability and things like that, that actually help the user, as opposed to competing on "but we're the only people who *really* know how.doc format works".
(IMO governments should use open formats; private companies shouldn't be required to use open formats, except when they interact with governments, which would hopefully mean they see the sense of going with open formats throughout.)
(Of course, if someone decides the best way to document their format is to open-source a sample implementation, that'd be nice too...)
Since everyone else is mentioning what worked for them...
I use a Psion Revo (a.k.a. Diamond Mako in the USA - it's a small 16MB palmtop with a keyboard, which might have been worth $200 a couple of years ago) as a diary. It's complete overkill for a student, but the screen's larger and clearer than the 2MB Psion 3c I used for 3 years before that (similar price when new), I can run Zork and Nethack on it if I want something to amuse myself, and the "busy" view in its agenda applet is wonderful for seeing at a glance how much time I have. (And, yes, I do find it better than a paper diary, if only because it's small enough that I always carry it.)
I take lecture notes with a pen and paper though. (I'm a maths student on a known-to-be-hard course; if you write down everything our lecturers say, you get a complete set of notes for the course, but it's pretty dense and you do need to write *everything*. Your course may vary.)
I'm planning to get a laptop at some point, possibly an iBook or Powerbook, but I doubt I'd take notes on it during lectures (I might give it a go in a course I don't care about, but I doubt I can type that fast); I might scan or even LaTeX my handwritten notes after writing them though, since if I do that, I'll be able to carry a complete set of notes around a lot more easily.
For general computing I've been happy with desktop computers (I have a quiet slow one as a fileserver and firewall, and a fast noisy one for games and general "workstation" use) but I'd prefer a "workstation" that was actually quiet, which is the main reason I'm considering a laptop.
In the UK we have alphanumeric postcodes which look something like NE12 3AB. The 1 or 2 letters at the beginning are loosely mnemonic (NE is the area around Newcastle), the 1 or 2 digits complete the code for a region, then the last digit and two more letters identify the street (a postcode in a residential area usually covers about 10 houses, so to uniquely identify an address, you need a postcode and a house number).
Or do we? That's not actually the Royal Mail's preferred format any more; they give discounts on bulk mail (by which I mean things like phone bills which are sent out in huge batches, not just junk mail) if it's marked with a "mailsort" code and received in the correct order. The mailsort code is a US-style purely numeric identifier, because it's easier to do OCR on something you know has to be a number - numbers are much less ambiguous than letters.
In a way we have Internet-like hierarchical distributed routing, I suppose (and I'm sure this is something that inspired early Internet development): if (outside the UK) you write
123 High Street Somewhere Some City NE12 3AB United Kingdom
on a letter, your local post office just needs to understand the last line in order to route it correctly. I'm sure other countries' conventions work similarly (although I seem to remember Russians write the address in "big-endian" order, with the country and city at the top, and the street and number at the bottom).
Presumably they're self-playing (like a self-extracting zip file, but for video) so your recipient just needs the same OS and not the same player. I've seen video encoders that offer this (RAD Game Tools used to, I think).
I'm not saying I think it's a good idea (I think it's a very bad idea, actually) but if all the recipients have Windows on x86, it's convenient and doesn't require much clue.
The BBC do have the advantage that they're a government-funded public service, rather than a profit-making business. In the worst case, that'd make them as bad as you describe, but they seem to have avoided that.
(They do have pretty serious competition on mainstream TV from the commercial ITV and Channel 4, although commercial radio is bad enough here that the BBC wins by default)
They're often rather critical of the government, actually, and in many disputes they're accused of being biased by both sides, which might well mean they're uncomfortably close to being balanced.
It's amusing to see the grandparent post commenting on Radio 1's larger playlists though, since some of the Radio 1 DJs have been known to complain (subtly, of course) about the commercial crap they're made to play. I hate to think what Clear Channel must be like if that's an improvement:-)
The computing service (who're responsible for the university and student networks) monitor general levels of traffic; if you've been using a lot of bandwidth for extended periods of time, they'll contact you, ask you what your excuse is, and tell you to slow down. The idea is that after a few warnings they'll disconnect your network socket, but most people take the hint.
Just looking at the stats rather than the protocol is also good for plausible deniability, since they don't particularly want to know the specifics of illegal file sharing and the like; they have been known to specifically stop a Direct Connect hub, but IIRC that was after another student had a private feud with the hub operator and decided to report them, after which the computing service had little choice.
They also occasionally scan random IPs for common server and trojan ports, then connect to some servers to see what banners etc. they produces, but this is more an anti-h4x0r thing than anything else; they don't even seem to mind students running low-traffic web servers on port 80, but they're likely to contact the student and verbally cluebat them if the server says it's IIS.
This isn't PNG 2.0, it's PNG second edition: think of it as being like the difference between Win98 and Win98 second edition.
When the W3C release a "second edition" recommendation, it's mostly editorial changes - see the changes summary in the PNG recommendation (or see XML 1.0 second edition, which is the current XML spec, for an example of another "second edition").
The linked spec is basically compatible with the original version, but some of the conditions for conformance have been tightened up (not that that matters for IE purposes since it didn't conform anyway) and the necessary verbiage to use the text as an ISO standard has been added (W3C policy is to release "recommendations" which are treated like standards, but this one is actually going to be a standard in theory as well as in practice).
Slashdot readers are a pretty small proportion of the population, you know. If someone's posting goatse links or trolls as Fred Smith rather than as Anonymous Coward, that doesn't really help you, and even if you get pissed off enough to work out where they live, what's the chance that you'll be close enough, and pissed-off enough, to do anything?
I've seen technical mailing lists that enforce "plausible names" though, those are reasonable in some contexts. (They won't let you subscribe unless you either give your real name, or invent a plausible "real name" - they basically want to know which user they're talking to, even if they don't actually have a real name.)
I often post under my real name myself (well, my real initials, and my real name is readily available if you have half a clue where to look), but I think I should be able to post/etc. under an alias, the same way I could claim my name was Fred Smith, or just not tell you, if I met you in real life. Of course, as in real life, you'd be within your rights not to listen to me, but isn't that is why Slashdot has a "filter ACs" option?
Well, share is for platform-independent data, so that's out. (A Mac and a PC with the same Linux distro and packages should be able to use the same NFS-mounted/usr/share tree, hence the name "share"; this matters more on traditional Unix hardware than it does now).
The rest are all possibilities, depending on whether you or your distribution vendor installed Mozilla, and whether you or they consider Mozilla to be a monolithic "black box" (like Windows apps) or an integrated part of the system (so it's easy for Galeon or other Gecko-based browsers to embed it).
It's valuable to have/usr/local and/usr separate - that way you, the local sysadmin (installing self-compiled stuff to/usr/local) and your package management system (installing to/usr) will never get in each others' way./opt vs./usr/local is a bit more subtle - you're meant to use/opt for identifiable "modules" which could be removed without side effects (I use it for games), and/usr/local for things which fit into the traditional Unix hierarchy (if you installed Mozilla in/opt the executable should be something like/opt/mozilla/mozilla or/opt/mozilla/bin/mozilla, if you installed in/usr/local it should just be/usr/local/bin/mozilla). Some distros don't have even have a/opt directory in the default install (Debian doesn't).
I realise it's not ideal, particularly with some of the more subtle points (share vs. lib,/usr/local vs./opt), but it's pretty much standardized by now.
(I wish all my dotfiles followed a similar hierarchy, actually - I've started using symlinks to get the caches in ~/.tmp and the important config files in ~/.etc, so I can leave out.tmp when I do backups)
Some of the merging Gobolinux does seems like overkill; for instance, the benefit of having the/usr hierarchy is that you can put all the critical system files (/bin,/lib,...) on a separate, smaller partition, which can sometimes even be read-only, guaranteeing that you have a bootable system.
You can only apply a GPL-like license to something on which you hold the copyright.
When you GPL that program, you're saying "I, the author, hereby give you permission to [...] as long as you [...]"; if parts of the program aren't your property, that permission isn't yours to give, so it's meaningless.
(Counter-example: If you gave me permission to distribute copies of Windows, and I did, I'd be breaking the law, because Microsoft own Windows and you don't. If *Microsoft* gave me permission to distribute copies of Windows, I'd be able to do so legally.)
Diceware looks like a nice way to generate random yet fairly memorable passphrases, for people whose typing is better than their memory. All you need is a list of 6^5 memorable words or almost-words, like the two English lists provided on that website (they've included almost-words like aaaa and 123 as well as real words, to keep the average word length down). Roll 5 dice (5D6 for roleplayers/Warhammer players:-) and pick a word from the list. Repeat until you have a strong passphrase.
Assuming an attacker knows you used Diceware and has a copy of the word list you used, a 5-word passphrase chosen like this is about as hard to brute-force as 64-bit encryption, and a 10-word passphrase is about as strong as the 128-bit symmetric encryption component of PGP.
It's handy, as my eyesight isn't the best and spending 12 hours a day in front of a monitor will -really- stress them out. A quick mutter of "damn designers" when I hit a page that's forcing small font sizes...
I can just imagine some future speech recognition system where saying "damn designers" to a web browser will switch off animation and enable sensible fonts:-)
I've actually used this sort of exception-handling at work, for an unexpected but not fatal condition in a program where next and last weren't suitable. You don't need $SIG{__DIE__} hooks, and in fact they're officially frowned upon by the Perl 5.x developers - instead, you wrap the whole thing in an "eval", which catches exceptions, then test a special "error" variable which contains the exception, or a false value if there wasn't one.
Catching all exceptions is bad (I still wanted my program to die on I/O errors), so you have to be a bit selective.
eval { # this is like "try"
while()
{...
if($need_to_terminate)
{ # this is like "throw"
die "Terminating loop";
}....
}
}
# this next bit is like "catch" # I forget exactly what the punctuation is; I *think* it's $? you need to look at though if($?) {
if($? eq "Terminating loop")
{
warn "Non-fatal exception, carrying on";
# or whatever else you want to do
} else {
die; # re-throw the exception to let someone else handle it
} }
The exceptions can be objects instead of strings, too.
I've also used this mechanism for a sort of more user-oriented "stack trace": the parse_line function (which doesn't know where its data came from) dies on errors, with a message which includes the offending line of a file, while the parse_file function catches exceptions provided by parse_line, adds something like "Error was in/home/smcv/foo.log line 345" to the error message, and re-throws the exception.
(It's generally useless to users of a logfile-mangling program to see where the error happened in your code, unless they start debugging your program, in which case they can turn on Carp for themselves; what's useful is to see where the error was in the format of their input file.)
I'm at Cambridge University, where students get a username consisting of their initials plus a (by now fairly large) sequential number (so John Michael Smith might be jms112), which tends to be a lot more memorable (only 3 arbitrary digits to remember). Students use these as their e-mail addresses and to log in to all lab or library computers, so they're easy to remember from frequent use as well.
This userID also appears on the college food/rent bill, so I assume they're just using these usernames as the unique identifiers in their billing system (which seems to make sense; after all, you're going to have a slightly memorable username generated for you anyway).
The other part of the solution is data protection law - as with the parent post, this is a European Union law, meaning it applies to governments and not to individuals (the EU law is like a specification, EU governments have to impose laws that "implement" it).
Google for "Data Protection Act 2000" to see the UK's version, which came into force in 2001 (an earlier version, which was a bit less restrictive, has been around since the early 90s) and places specific restrictions on storage and use of "personally identifiable information". As far as I can see, this makes trading e-mail lists without specific permission illegal - it also has other nice effects, like preventing companies from sharing any other information they've gathered about you without your permission. (So your bank can't tell any other companies about your spending without your permission, for instance.)
The USA doesn't have a similar law, which actually means companies in Europe aren't allowed to send personal data to their American branch without the subject's permission.
If AOL, Microsoft, et al won't do it, I bet some intrepid programming brains will write "switchboard" type server software that will do it for them, assuming it hasn't been done already.
It's been done. There are two approaches:
1) Multi-network clients, of which the most popular are probably Trillian (Windows) and Gaim (Unix/Gtk+). These open separate connections to AIM, ICQ, MSN, Yahoo, and so on, at the same time - you need accounts on the appropriate services first though. In particular, you need both an AIM and an ICQ account if you want to talk to people on both AIM and ICQ, despite the fact that they use the same servers and the same protocol - the only difference these days seems to be that ICQ has numbers, AIM has screennames, and the server won't let either sort of user talk to the other.
2) Jabber [http://www.jabber.org]. As well as Jabber servers themselves (Yet Another messaging protocol, this one based on XML and running on an open-source server), this has a nifty feature where you provide your Jabber client with your username and password for "foreign" services, the client passes them on to the server, and the server logs on to AIM/etc. as you, converting incoming messages to Jabber messages to send to your client (sort of like the webmail services which offer to fetch your POP3 mail into the webmail account so it's all in one place, but for IM instead of mail). For a while, AOL IP-banned the main public Jabber server from AIM - they obviously weren't happy about the idea.
If you ran a search engine, wouldn't you want to block other search engines (and indeed your own search engine) from indexing your results? Search engine results pages probably shouldn't appear as hits on search engine results pages, after all:-)
(Google haven't blocked/services/ and so on, so all their static content is indexable - it's just the results that aren't)
Precisely. Smaller sites can afford a domain (.uk domains are only about the equivalent of US$10 a year) but not a SSL cert (around US$900, unless you want a 40-bit false sense of security). If browsers had some sort of web of trust model, perhaps SSL would be feasible, but with the current certifying authority model, the whole thing's very reliant on the good will of the CAs (who obviously have an interest in keeping prices high).
SSL also breaks HTTP/1.1 name-based virtual hosting, since you have to figure out which certificate to use before the client has sent you a Host: header (IOW, each physical computer can only host one SSL hostname).
Slashdot Mirror
Programs compiled with gcc don't have to be GPLed. If they did, Apple wouldn't be using it in Mac OS X (the Darwin kernel is open-source but not GPL-compatible, the high-level bits like Aqua are proprietary.)
The community around Polycount tends to be strictly opposed to plagiarism, and it seems to a good idea to ask authors' permission before doing anything to their model (as well as being courteous - they like to know that their models are appreciated :-)
Some models on Polycount don't actually have a permissions statement, meaning that if you want to be picky, it may not even be legal for Polycount to distribute them. Some have non-commercial clauses in the readme, or other encumberances.
Some Q3 models use, or are based on, Quake 3 stock animations - I'm not sure about the copyright status of these, and whether they're a large enough proportion that the model counts as a derived work.
As for "Free" rather than "free", I haven't seen a model on Polycount yet that's freely modifiable, or meets the FSF or OSI guidelines for free software/open source software - so if you want open-source code combined with unmodifiable graphics, that's fine, but if you want your artwork to be licensed in the same way as your code, Polycount is unlikely to help you unless you specifically ask the authors for more permission.
Forcing one particular format is a recipe for disaster - as soon as a feature it doesn't support comes along, you need a new law.
What is worth forcing is that the format is documented sufficiently well that it can be reimplemented.
In this particular case, the format is easy to document, because you can just say "It's XML, it conforms to specifications A, B and C, and the whole thing is in a zip file as described by RFC 1234" - so all you actually need to document is the particular "application" (sub-language) of XML that you're using, preferably along with forwards-compatibility notes like "unknown elements are a fatal error" or "unknown attributes must be ignored" or whatever you like.
If you have a law forcing data to be in properly-documented formats, proprietary software companies are free to implement apps that work with that format, without open-sourcing anything.
.doc format works".
They'll just have to compete (with open-source apps, and with other proprietary apps) on features, ease of use, reliability and things like that, that actually help the user, as opposed to competing on "but we're the only people who *really* know how
(IMO governments should use open formats; private companies shouldn't be required to use open formats, except when they interact with governments, which would hopefully mean they see the sense of going with open formats throughout.)
(Of course, if someone decides the best way to document their format is to open-source a sample implementation, that'd be nice too...)
Since everyone else is mentioning what worked for them...
I use a Psion Revo (a.k.a. Diamond Mako in the USA - it's a small 16MB palmtop with a keyboard, which might have been worth $200 a couple of years ago) as a diary. It's complete overkill for a student, but the screen's larger and clearer than the 2MB Psion 3c I used for 3 years before that (similar price when new), I can run Zork and Nethack on it if I want something to amuse myself, and the "busy" view in its agenda applet is wonderful for seeing at a glance how much time I have. (And, yes, I do find it better than a paper diary, if only because it's small enough that I always carry it.)
I take lecture notes with a pen and paper though. (I'm a maths student on a known-to-be-hard course; if you write down everything our lecturers say, you get a complete set of notes for the course, but it's pretty dense and you do need to write *everything*. Your course may vary.)
I'm planning to get a laptop at some point, possibly an iBook or Powerbook, but I doubt I'd take notes on it during lectures (I might give it a go in a course I don't care about, but I doubt I can type that fast); I might scan or even LaTeX my handwritten notes after writing them though, since if I do that, I'll be able to carry a complete set of notes around a lot more easily.
For general computing I've been happy with desktop computers (I have a quiet slow one as a fileserver and firewall, and a fast noisy one for games and general "workstation" use) but I'd prefer a "workstation" that was actually quiet, which is the main reason I'm considering a laptop.
In the UK we have alphanumeric postcodes which look something like NE12 3AB. The 1 or 2 letters at the beginning are loosely mnemonic (NE is the area around Newcastle), the 1 or 2 digits complete the code for a region, then the last digit and two more letters identify the street (a postcode in a residential area usually covers about 10 houses, so to uniquely identify an address, you need a postcode and a house number).
Or do we? That's not actually the Royal Mail's preferred format any more; they give discounts on bulk mail (by which I mean things like phone bills which are sent out in huge batches, not just junk mail) if it's marked with a "mailsort" code and received in the correct order. The mailsort code is a US-style purely numeric identifier, because it's easier to do OCR on something you know has to be a number - numbers are much less ambiguous than letters.
In a way we have Internet-like hierarchical distributed routing, I suppose (and I'm sure this is something that inspired early Internet development): if (outside the UK) you write
123 High Street
Somewhere
Some City
NE12 3AB
United Kingdom
on a letter, your local post office just needs to understand the last line in order to route it correctly. I'm sure other countries' conventions work similarly (although I seem to remember Russians write the address in "big-endian" order, with the country and city at the top, and the street and number at the bottom).
Presumably they're self-playing (like a self-extracting zip file, but for video) so your recipient just needs the same OS and not the same player. I've seen video encoders that offer this (RAD Game Tools used to, I think).
I'm not saying I think it's a good idea (I think it's a very bad idea, actually) but if all the recipients have Windows on x86, it's convenient and doesn't require much clue.
The BBC do have the advantage that they're a government-funded public service, rather than a profit-making business. In the worst case, that'd make them as bad as you describe, but they seem to have avoided that.
:-)
(They do have pretty serious competition on mainstream TV from the commercial ITV and Channel 4, although commercial radio is bad enough here that the BBC wins by default)
They're often rather critical of the government, actually, and in many disputes they're accused of being biased by both sides, which might well mean they're uncomfortably close to being balanced.
It's amusing to see the grandparent post commenting on Radio 1's larger playlists though, since some of the Radio 1 DJs have been known to complain (subtly, of course) about the commercial crap they're made to play. I hate to think what Clear Channel must be like if that's an improvement
The computing service (who're responsible for the university and student networks) monitor general levels of traffic; if you've been using a lot of bandwidth for extended periods of time, they'll contact you, ask you what your excuse is, and tell you to slow down. The idea is that after a few warnings they'll disconnect your network socket, but most people take the hint.
Just looking at the stats rather than the protocol is also good for plausible deniability, since they don't particularly want to know the specifics of illegal file sharing and the like; they have been known to specifically stop a Direct Connect hub, but IIRC that was after another student had a private feud with the hub operator and decided to report them, after which the computing service had little choice.
They also occasionally scan random IPs for common server and trojan ports, then connect to some servers to see what banners etc. they produces, but this is more an anti-h4x0r thing than anything else; they don't even seem to mind students running low-traffic web servers on port 80, but they're likely to contact the student and verbally cluebat them if the server says it's IIS.
This isn't PNG 2.0, it's PNG second edition: think of it as being like the difference between Win98 and Win98 second edition.
When the W3C release a "second edition" recommendation, it's mostly editorial changes - see the changes summary in the PNG recommendation (or see XML 1.0 second edition, which is the current XML spec, for an example of another "second edition").
The linked spec is basically compatible with the original version, but some of the conditions for conformance have been tightened up (not that that matters for IE purposes since it didn't conform anyway) and the necessary verbiage to use the text as an ISO standard has been added (W3C policy is to release "recommendations" which are treated like standards, but this one is actually going to be a standard in theory as well as in practice).
Slashdot readers are a pretty small proportion of the population, you know. If someone's posting goatse links or trolls as Fred Smith rather than as Anonymous Coward, that doesn't really help you, and even if you get pissed off enough to work out where they live, what's the chance that you'll be close enough, and pissed-off enough, to do anything?
I've seen technical mailing lists that enforce "plausible names" though, those are reasonable in some contexts. (They won't let you subscribe unless you either give your real name, or invent a plausible "real name" - they basically want to know which user they're talking to, even if they don't actually have a real name.)
I often post under my real name myself (well, my real initials, and my real name is readily available if you have half a clue where to look), but I think I should be able to post/etc. under an alias, the same way I could claim my name was Fred Smith, or just not tell you, if I met you in real life. Of course, as in real life, you'd be within your rights not to listen to me, but isn't that is why Slashdot has a "filter ACs" option?
--SMcV
/usr/share/plugins /usr/share/netscape/plugins /usr/share/mozilla/plugins
/usr/share tree, hence the name "share"; this matters more on traditional Unix hardware than it does now).
/usr/local and /usr separate - that way you, the local sysadmin (installing self-compiled stuff to /usr/local) and your package management system (installing to /usr) will never get in each others' way. /opt vs. /usr/local is a bit more subtle - you're meant to use /opt for identifiable "modules" which could be removed without side effects (I use it for games), and /usr/local for things which fit into the traditional Unix hierarchy (if you installed Mozilla in /opt the executable should be something like /opt/mozilla/mozilla or /opt/mozilla/bin/mozilla, if you installed in /usr/local it should just be /usr/local/bin/mozilla). Some distros don't have even have a /opt directory in the default install (Debian doesn't).
/usr/local vs. /opt), but it's pretty much standardized by now.
.tmp when I do backups)
/usr hierarchy is that you can put all the critical system files (/bin, /lib, ...) on a separate, smaller partition, which can sometimes even be read-only, guaranteeing that you have a bootable system.
Well, share is for platform-independent data, so that's out. (A Mac and a PC with the same Linux distro and packages should be able to use the same NFS-mounted
The rest are all possibilities, depending on whether you or your distribution vendor installed Mozilla, and whether you or they consider Mozilla to be a monolithic "black box" (like Windows apps) or an integrated part of the system (so it's easy for Galeon or other Gecko-based browsers to embed it).
It's valuable to have
I realise it's not ideal, particularly with some of the more subtle points (share vs. lib,
(I wish all my dotfiles followed a similar hierarchy, actually - I've started using symlinks to get the caches in ~/.tmp and the important config files in ~/.etc, so I can leave out
Some of the merging Gobolinux does seems like overkill; for instance, the benefit of having the
You can only apply a GPL-like license to something on which you hold the copyright.
When you GPL that program, you're saying "I, the author, hereby give you permission to [...] as long as you [...]"; if parts of the program aren't your property, that permission isn't yours to give, so it's meaningless.
(Counter-example: If you gave me permission to distribute copies of Windows, and I did, I'd be breaking the law, because Microsoft own Windows and you don't. If *Microsoft* gave me permission to distribute copies of Windows, I'd be able to do so legally.)
Europe has data protection law to control who gets your personal information (click here for info about the UK's implementation). Shouldn't you have the same?
Diceware looks like a nice way to generate random yet fairly memorable passphrases, for people whose typing is better than their memory. All you need is a list of 6^5 memorable words or almost-words, like the two English lists provided on that website (they've included almost-words like aaaa and 123 as well as real words, to keep the average word length down). Roll 5 dice (5D6 for roleplayers/Warhammer players :-) and pick a word from the list. Repeat until you have a strong passphrase.
Assuming an attacker knows you used Diceware and has a copy of the word list you used, a 5-word passphrase chosen like this is about as hard to brute-force as 64-bit encryption, and a 10-word passphrase is about as strong as the 128-bit symmetric encryption component of PGP.
It's handy, as my eyesight isn't the best and spending 12 hours a day in front of a monitor will -really- stress them out. A quick mutter of "damn designers" when I hit a page that's forcing small font sizes...
:-)
I can just imagine some future speech recognition system where saying "damn designers" to a web browser will switch off animation and enable sensible fonts
I've actually used this sort of exception-handling at work, for an unexpected but not fatal condition in a program where next and last weren't suitable. You don't need $SIG{__DIE__} hooks, and in fact they're officially frowned upon by the Perl 5.x developers - instead, you wrap the whole thing in an "eval", which catches exceptions, then test a special "error" variable which contains the exception, or a false value if there wasn't one.
... ....
/home/smcv/foo.log line 345" to the error message, and re-throws the exception.
Catching all exceptions is bad (I still wanted my program to die on I/O errors), so you have to be a bit selective.
eval { # this is like "try"
while()
{
if($need_to_terminate)
{ # this is like "throw"
die "Terminating loop";
}
}
}
# this next bit is like "catch"
# I forget exactly what the punctuation is; I *think* it's $? you need to look at though
if($?)
{
if($? eq "Terminating loop")
{
warn "Non-fatal exception, carrying on";
# or whatever else you want to do
} else {
die; # re-throw the exception to let someone else handle it
}
}
The exceptions can be objects instead of strings, too.
I've also used this mechanism for a sort of more user-oriented "stack trace": the parse_line function (which doesn't know where its data came from) dies on errors, with a message which includes the offending line of a file, while the parse_file function catches exceptions provided by parse_line, adds something like "Error was in
(It's generally useless to users of a logfile-mangling program to see where the error happened in your code, unless they start debugging your program, in which case they can turn on Carp for themselves; what's useful is to see where the error was in the format of their input file.)
Is there a do/while statement in Python?
...
The usual idiom seems to be:
while 1:
if exitCondition:
break
Whenever /. has a crypto story, someone posts something like this:
there exists a known algorithm to factorise primes in polynomial time
which is perfectly true. Even better, it works in constant time:
def factor(prime):
return (prime, 1)
(translation into languages other than Python is left as an exercise to the reader)
Factorizing composites efficiently is how you break RSA and related cryptographic algorithms.
I'm at Cambridge University, where students get a username consisting of their initials plus a (by now fairly large) sequential number (so John Michael Smith might be jms112), which tends to be a lot more memorable (only 3 arbitrary digits to remember). Students use these as their e-mail addresses and to log in to all lab or library computers, so they're easy to remember from frequent use as well.
This userID also appears on the college food/rent bill, so I assume they're just using these usernames as the unique identifiers in their billing system (which seems to make sense; after all, you're going to have a slightly memorable username generated for you anyway).
The other part of the solution is data protection law - as with the parent post, this is a European Union law, meaning it applies to governments and not to individuals (the EU law is like a specification, EU governments have to impose laws that "implement" it).
Google for "Data Protection Act 2000" to see the UK's version, which came into force in 2001 (an earlier version, which was a bit less restrictive, has been around since the early 90s) and places specific restrictions on storage and use of "personally identifiable information". As far as I can see, this makes trading e-mail lists without specific permission illegal - it also has other nice effects, like preventing companies from sharing any other information they've gathered about you without your permission. (So your bank can't tell any other companies about your spending without your permission, for instance.)
The USA doesn't have a similar law, which actually means companies in Europe aren't allowed to send personal data to their American branch without the subject's permission.
If AOL, Microsoft, et al won't do it, I bet some intrepid programming brains will write "switchboard" type server software that will do it for them, assuming it hasn't been done already.
It's been done. There are two approaches:
1) Multi-network clients, of which the most popular are probably Trillian (Windows) and Gaim (Unix/Gtk+). These open separate connections to AIM, ICQ, MSN, Yahoo, and so on, at the same time - you need accounts on the appropriate services first though. In particular, you need both an AIM and an ICQ account if you want to talk to people on both AIM and ICQ, despite the fact that they use the same servers and the same protocol - the only difference these days seems to be that ICQ has numbers, AIM has screennames, and the server won't let either sort of user talk to the other.
2) Jabber [http://www.jabber.org]. As well as Jabber servers themselves (Yet Another messaging protocol, this one based on XML and running on an open-source server), this has a nifty feature where you provide your Jabber client with your username and password for "foreign" services, the client passes them on to the server, and the server logs on to AIM/etc. as you, converting incoming messages to Jabber messages to send to your client (sort of like the webmail services which offer to fetch your POP3 mail into the webmail account so it's all in one place, but for IM instead of mail). For a while, AOL IP-banned the main public Jabber server from AIM - they obviously weren't happy about the idea.
If you ran a search engine, wouldn't you want to block other search engines (and indeed your own search engine) from indexing your results? Search engine results pages probably shouldn't appear as hits on search engine results pages, after all :-)
/services/ and so on, so all their static content is indexable - it's just the results that aren't)
(Google haven't blocked
the fee for a certificate
Precisely. Smaller sites can afford a domain (.uk domains are only about the equivalent of US$10 a year) but not a SSL cert (around US$900, unless you want a 40-bit false sense of security). If browsers had some sort of web of trust model, perhaps SSL would be feasible, but with the current certifying authority model, the whole thing's very reliant on the good will of the CAs (who obviously have an interest in keeping prices high).
SSL also breaks HTTP/1.1 name-based virtual hosting, since you have to figure out which certificate to use before the client has sent you a Host: header (IOW, each physical computer can only host one SSL hostname).
Almost right, but /dev/null produces no data when read. You want /dev/zero, which produces an infinite stream of zeros.