I've found that bringing in alien paradigms can be effective. For example, if you're writing in C, simulate functional programming mechanisms, such as closures. This by itself is not enough, because someone could be reading the code who knows Scheme, but it is nonetheless effective, because it will throw a significant portion of your audience into unmitigated confusion. Only people familiar with the alien paradigm will even have a *chance* at understanding what the code is doing.
> Anyhow, I was perusing the code. It was pretty sloppy, one-letter variable > names, multiple statements crammed together on single lines.
Done correctly, multiple statements on a line can actually make code easier to read, by allowing more of the context to fit on the screen at once. It's harder to defend one-letter variable names, except perhaps for very tightly scoped lexicals, where you can see every reference to the variable in a small handful of lines.
> So I get to this one section of code. I can't even remember how it worked > now, but it was this convoluded for() statement that flipped a flag and > did some weird ass computation. It took me about 10 minutes of "stepping > through" it in my head to figure out what it was doing.
Heh. C-style for loops are a pet peave of mine. Fortunately, they're going the way of the dodo in Perl6, and all for loops will really be foreach loops. (Oh, you're still programming in C? I'm sorry.)
> It was calculating leap-years. I actually stared at it in shock, imagining > how much time and energy this kid spent figuring out the worlds most > assinine way to figure out if it's leapyear. I would have just wrote > "if (year%4 == 0) { days_in_feb=29; }" or something of the sort.
That would get the answer wrong sometimes. I would write something like use DateTime; if (DateTime->now->is_leap_year()) { whatever() }
> The fact of the matter is that once you leave your PC unattended, no > amount security is going to save you.
There are some things you can do, they're just a much bigger pain than locking up the system in a room with tigers and cobras and security cameras. For example, you can use an encrypted filesystem that requires a long key to be entered on the keyboard at boot time. That leaves you open to a sniffing attack with a KeyGhost or similar device, or clever placement of a hidden camera to record your typing in the key. There are ways to guard against that too, by using a challenge and response type method to unlock the encrypted filesystem, but that can get very complex and require quite a lot out of the user. (Of course, if you don't reboot very often, that would help minimize the annoyance factor... but then you don't use the authentication system very often, so you're very liable to forget it, at which point you now have to write it down and lock that up in a safe...) This amounts to quite a lot of hassle, and it really would be simpler to lock the system in a room with dogs and guards.
Additionally, all of this stuff still allows the user to boot your hardware (albeit not your software) and probably get onto the network, though, so then you have to harden your network against internal attacks, which means careful firewalling and stuff. Which, admittedly, you theoretically should do anyway, in principle, but I suspect most of us don't.
> the PC in question needs to be set up to boot from USB
Actually, I just noticed that it comes with a bootable mini-CD to get around that issue. I suspect you'd end up using that most of the time. Does anyone know whether the CD then boots the USB drive as the root partition, so that you can function as if you'd booted from it directly, once you're booted? If it does, then that would give it some advantages over other LiveCD + USB drive options (e.g. MandrakeMove), since you could e.g. easily install upgrades.
> Perhaps it's a credible alternative to a remote login to your main computer. > You are still going to need to find an existing computer to plug this thing > into though.
That wouldn't be a problem in North America (when was the last time you were more than three rooms away from a computer, really?), except for the fact that the PC in question needs to be set up to boot from USB, which narrows down your options rather a lot, since approximately zero percent of the installed base of PCs out there are set up to do that. It's not even an *option* on most BIOSes, and one imagines that even on systems with the capability the default would still be to boot via a more traditional source, such as the first floppy drive, or a hard drive or ATAPI CD-ROM drive on one of the IDE channels.
Many modern systems have, in theory, the capability to boot off the network -- but it's seldom the default. You could carry around a smallish LTSP server and a crossover cable, but plugging it into most systems wouldn't get you anywhere, because they're not set up to boot that way.
Of course, if you have unobserved physical access to the system you could *alter* the way it's set up, but at that point this ceases to be a benign hey-can-I-borrow-your-hardware-for-a-few-m oments thing and becomes rather more invasive. And there are still a lot of systems that don't even have the capability to boot from USB.
It's an interesting concept, but the LiveCD + portable data drive solution (a la MandrakeMove) seems more realistically useful at this point.
> If the paper is turned in as a printed document, the teacher will have > to re-type the entire thing and Google each part for potential matches.
You're assuming we need to catch all the students who merely plagiarize small bits and pieces, _basically_ writing the bulk of their own paper, or at least putting it together, but without doing proper citations of their sources.
If all we want to do is catch people who copy most or all of the paper pretty much verbatim, then we don't have to search for each and every part of the paper; it's enough to look through the paper, pick out a couple of long phrases that seem like they would be reasonably close to unique (i.e., not something like "it has been discovered that", which would turn up lots of false hits), and do phrase searches for them. Any matches that you find, you check to see if more of the paper matches that source than just that phrase.
That's the good news. The bad news is, Google's only going to find you mostly stuff that's readily available on the web. If the student copies a Wikipedia article mostly-verbatim, this will find it. But there's quite a lot of stuff out there that Google doesn't index. If the student uses Electric Library or EbscoHost or SIRS to find a journal article and turns *that* in as their paper, then finding the student's source is not quite as simple as a single Google search. It can still be done, but it's harder.
> > One last thing though - when (if ever) will Mozilla mail change away from > > using.mbx/mailbox files and move to something like what Sylpheed uses > > (1 file per email). > > Wouldn't that create a ton of overhead with searches
It would also create a ton of flexibility with searches. (You can e.g. easily look for messages stored in a given folder within a certain date range that contain a certain email address in the From: field and also contain a certain word in the body -- without doing hardly any parsing.) However, Mozilla Mail has never really been about flexibility; it's more about simplicity. People who want flexibility spend the three weeks of initial learning overhead so they can use Gnus, at which point they can do pretty much anything -- and even more than that if they know some lisp. Mozilla Mail is for people who just want to start reading mail right away, instead of reading a lot of info pages. (Yeah, I use Gnus. But I use a mozilla.org browser, and I'm experimenting with Sunbird.)
> I think they need to use different animals for each program though. > Firefox, thunderBird, Why not something like aquaMoose?
SunMarmot? Too dorky. SunShrew? Too grouchy. SunTzu? Too obscure, and potentially offensive. SunWoodchuck? Too long. (Just SunChuck perhaps?) SunDeer? Closer, but I don't like a product name ending in r; it's too easy to get sloppy on the pronunciation. SunShark? Too toothy. SunGorilla? Overpowered. SunSquirrel? Nuts. SunSwift? But that's a bird. SunMouse? Too timid. SunDog? Too ghetto. SunSloth? Too lazy, especially for a calendar app -- people would be afraid of missing appointments. SunCat? Too many people don't like cats. SunTiger? We're not Apple. SunLynx? There's already a browser called Lynx. And so on.
There are some possibilities, though. Personally, for a calendar, fitting in with the Firefox/Thunderbird theme, I like Stormcrow, because it warns you of things. Then again, if it were me, phoenix/firebird/firefox would be named something more original, like FlameThrush, and Thunderbird would be named more originally too, like BoomSwift (notice, another kind of bird), and then the calendar would have different lines to work along and could be something like DayOwl or FlashMartin or, indeed, StormCrow.
Linux isn't an OS at all; it's a kernel. Nevertheless, it was developed first and foremost for desktop use (albeit, for exteme geek desktop use). It had to develop for several months before it approached server usability, and for quite a while before it was really considered production-quality for servers. Even today, it is arguably not as well suited for servers as some other OSes (most particularly BSD), but it's a great desktop OS (for geeks -- though the extremeness of geekhood involved has declined substantially; it's no longer even really necessary to know C).
It's because *nobody* can stand Real. Their software has more nag screens than anything else you can download, and, additionally, it's feature-poor, doesn't work well, and uses (surprise) its own proprietary format in an attempt to lock people in. If a decent company with a useable product had reverse-engineered Apple's DRM system, public opinion might have been rather different, but Apple has a good reputation, and Real has such an extremely, utterly bad reputation, they make Microsoft look like a the poster boy for popular companies.
> Usenet isn't just any old NNTP connection, it's a public, widely used > network of NNTP and UUCP and whatever else connections
I'll go for usenet being a proper noun, but it's lowercased anyway because it derives etymologically from Unix culture and so is case-sensitive.
> The Internet is the same- it's not just a bunch of machines connected > together via TCP/IP, it's the collection of machines connected together > via TCP/IP.
The internet we have happens to be the only worldwide internetwork, but it would certainly be possible to have another. Imagine for a moment that during the third patent war in 2137, all interoperability between open-source systems and commercial systems is lost at the IPv8 level. Voila, two internets. In the more immediate present, imagine an internet of IPv6 hosts, and an internet of IPv4 hosts. Hosts that support both protocols would be on both internets. Or, if you're into sci-fi, imagine aliens that have their own internet, based on their own technology and protocols, and negotiations (and, if you prefer to imagine it, war) over how to get the two internets joined.
There's nothing inherently unique or singular about the concept of an internet; we call it _the_ internet merely because we only have one, just as in a small community you find people speaking of the library, the lumber yard, and so on. I have not yet lived in a community small enough to speak of "the restaurant".
> a LAN is a net, but not the Net
A LAN is a network, but it is not the only network -- but in a given place it might be the only one under consideration; at work, for example, we only have one network, and I regularly speak of "the network" at work, when I mean, "our LAN". It isn't whether there's only one or not that makes a thing a proper noun. If that were the case, universe would be a proper noun (since, by definition (everything that exists), there can only be one) but it's not, because the term refers to a concept (i.e., is a common noun), even though in practice there turns out to be just one of them. On the other hand, there might be *lots* of people named Bob Smith, but it's still a proper noun (or a proper noun phrase).
What next, will they also stop capitalizing "usenet" and "web browser" and "service provider" and "post office" and "bank" and "grep" and "sort"?
If internet were a proper noun, there would be other worldwide communications networks with other names. (No, don't say the phone network; phones and web browsers transmit their data over the *same* network, they just do it rather differently.) TCP/IP over avian carriers never really caught on, for some strange reason, so we only have one internet. But if we had did have two distinct internetworks, they'd both be internets.
usenet is an edge case -- it really probably ought to be a proper noun, but it comes out of Unix culture, so it's lowercase as a matter of case-sensitive spelling, even at the beginning of a sentence, like grep and sort (when sort is used as a proper noun -- when sort is used as a verb or a common noun it's from standard English and is capitalized according to the normal rules).
And yes, this is consistent with the normal rules of English, in the sense that the normal rules of English allow for exceptions based on sepcial rules pertaining to a given etymological source. There are also many English words that are CamelCased -- and I don't just mean computer words, either -- because of their etymology or the particulars of the field they come from. Similarly, words derived from foreign languages often form their plurals specially or are pronounced specially according to the rules of the source language or field, e.g., pianos and filet, respectively. case-sensitive spellings from Unix culture are consistent with this.
> I've never known a library to use 19" monitors for anything
We are moving in that direction. 19" CRTs (18" viewable) or 17" LCD, depending on physical space requirements. We quit buying 15" CRTs several years ago and are now choosing 19" over 17" a significant proportion of the time.
A 19" CRT these days barely costs any more than a 17" CRT, and they are significantly easier on the eyes of many of our patrons, who do not all have the eyesight they used to have. (Seems to be a bigger problem than when I was a kid. I blame it on television; don't know whether there's actually a cause-effect relationship there, but it feels good to blame television for something more concrete than moral degradation.) We find that going to 19" allows us to set the resolution to 1024x768 without making most folks squint; on a 17" monitor, 800x600 is pushing it for many and 1024x768 is right out; on a 15" monitor (of which we no longer have any except all-in-one units such as iMacs[1]) the resolution really can't be set any higher than 640x480, which isn't even an option on newer versions of certain OSes.
LCDs would be slightly better due to the extra inch you get out of their more honest screen size notation, except that they look like exrement at normal resolutions, basically forcing you to crank them up to their "native" res, which results in a lot of squinting. But we nevertheless use them sometimes, particularly in cramped locations where a 19" CRT would be unweildy. I look for LCDs with a relatively low ratio of native res to diagonal measure, to minimize squinting. (This is counterintuitive, since looking for smaller res numbers violates the "more is better" principle. I guess in this case the normally-inane "less is more" principle actually applies. Phooey.)
[1] And VT510s, which are even *smaller*, but those are amber-on-black, which
significantly cuts down on eyestrain right there, and in any case we'll
be moving away from them circa 2005-2006 when we migrate from Galaxy to
a more modern ILS.
> That'll be a couple of SSH terminal sessions and some FTPing then
Actually, I'm thinking one Cygwin window containing an X session, which you use with X11 forwarding to display the apps running on your *nix system. Oh, wait, you could just use *nix then. Remind me what I want XPSE for? Oh, yeah, because I wouldn't want people thinking I'm a computer geek who uses a weird custom OS that didn't come with the computer, now would I?
You will not believe how bad The Creeping Terror is until you watch it yourself. I cannot describe its badness (not that that's going to stop me from trying...) It makes the special effects in The Blob look stunning by comparison. It makes the acting in The Crawling Hand look like Academy Award material. It makes the character development in War of the Worlds look deep and meaningful. It makes the plotline of Spy Kids 2 seem almost plausible. It makes the camera work in Return of the Killer Tomatoes look impeccably professional. It is by far and away the most amazingly bad movie ever made. From the concept to the implementation, from the writing to the scenery, from the acting to the directing, it had *better* be the most impressively pathetic movie, because if there is a worse one, I think I will wet my pants when I find out about it.
In a nutshell, The Creeping Terror is about these alien things that come out of this landed spaceship. They look remarkably like people under a blanket crawling around, only much slower. Occasionally you can see shoes under the edges. Whenever these creeping things encounter anyone, the encountered party screams and freezes, because if they ran (or walked, or indeed crawled) away, the movie wouldn't have a plot. So they freeze, and scream, and wait for the terrible creeping blanket to come right up to them, at which point they have to crawl up into the thing to be eaten. This happens a whole bunch of times, while scientists and police and whatnot poke around the spaceship in a completely vain attempt to figure out what's going on. Eventually they find the first one (which got out of the spaceship when it landed) and figure out how to disable it, and then they inadvertently unleash the second one (which was up to then restrained in the ship), and go through the whole big thing *again*, and then when they disable the second one, the lead scientist realizes that it sent a signal back to the ship -- so they go back to the ship and try to shut it down before it transmits, but they fail, so it transmits (something) into space, causing the lead scientist to conclude that someday more of these things will come, and we'd better be ready, and then, mercifully, the movie ends.
And if you think that sounds long and boring and lame, that's the executive summary. Remember how I said the movie is so bad, I cannot describe its badness? Yeah, I meant that. I've barely scratched the surface. Don't believe me? You must watch the movie. It will exceed all your expectations of badness and leave you stunned, in denial that any movie could be so bad.
According to my analysis, only about 50% of spam is written in a Latin-derived character set[1], and that includes all that Spanish spam from Central and South America, all the money scams written in mostly ALL CAPS that are generally believed to come from Africa, and anything that comes from Western Europe or Australia (not much from there I think), and any spam written in English coming from Asia, which is doubtless no small amount, in *addition* to everything from the US.
*Most* of the other 50% is written in decidedly Asian character sets (though there's a little Cyrillic spam and the occasional Greek message). As much as 25% of all spam is in the GB2312 character set alone. I have a really hard time believing that's coming mostly from the US.
Maybe my sample is skewed (admittedly, I'm looking mostly at spam sent to one address), but it's a pretty good-sized sample (1.3GB).
[1] For UTF-8 messages, I base my "character set" data on the actual
characters used, which in spam are usually ideographic or syllabic
characters from Asia; most English-language spam is written in ASCII
or Latin-1 or something similar. Most English-language UTF-8 messages
are from mailing lists related to the open-source community. I suppose
this is probably because US localized builds of Outlook and its ilk
don't send unicode by default.
> Most likely it will come with something new and fancy called Windows
Yeah, I used to use that. But even then, I usually kept Emacs filling the whole width of the screen. I find that while it's useful to position two or three or even four windows vertically above one another, horizontal tiling is of less utility than the ability to work with wider content.
> Or you could take 20 mins to figure out the security settings you need on > an XP box to stop the user from installing thier own programs.
Right, so then you'd only have to worry about the apps that come with Windows, such as MSIE and MSOE. Fortunately those apps are very secure, as we all know, so the whole solution would be very robust... or something.
> I'm pretty sure this disqualifies me from being a philosopher
No, you can be a philosopher without drinking alcohol if you have something else you drink instead -- say, coffee or tea -- or some other quirky behavior that causes other people to write you off as eccentric. (This is a key difference between a philosopher and a thinker of some other sort, such as an analyst or a strategist: people can respect an analyst or strategist, because they believe his thinking has value; they look askance at a philosopher, because they believe his thinking is off the wall.)
> I know there is MacOS X, but that's kind of randomly chosen name (more > than a version number), no?
Lesse, there was System 6, System 7, MacOS 8, MacOS 9, then Mac OS X, which was at first 10.0 then 10.1 then 10.2 then 10.3 and soon 10.4. Looks like a number to me.
Nor was Mac the first OS to hit double-digit version numbers, and I think Solaris was there before Mandrake. Heck, Emacs is at version 21 now:-)
I did try Gentoo, and the experience was informative, but ultimately I came back to Mandrake. It's slightly more of a pain with Mandrake to install bleeding edge stuff (except for major apps like Mozilla, which are both easy and straightforward), and you spend slightly more time hunting down where the distribution puts things, but I found that for me, Gentoo outweighed its advantages in those areas with some disadvantages of its own.
Foremost, things just aren't as well integrated with Gentoo, in terms of the way things are set up. I think this mostly has to do with configuration -- sure,you have to reconfigure things with any distro to match your preferences, but with distros like Mandrake, the initial state is that the various components are all set up to interact well with one another. With Gentoo, I had some trouble getting things to play nicely with one another.
I expected Gentoo to solve dependency hell for me; it doesn't. In more than one case I had to manually track down packages that were causing emerge to fail, and individually install (or uninstall!) them. In one case I had to hunt all over the internet and eventually find that in order to emerge world I had to first uninstall certain core Perl modules. That should never be. This is different from the situation in Mandrake, but IMO it's not better.
It was also not always obvious what package I had to install to get a certain thing. I had an horrific time figuring out how to install the themes for Gnome. With Mandrake, this is not a problem.
Mandrake has its problems, sure. It's not the most secure distro, and there can be problems if you only want to upgrade some components and not others. (For example, if you want to keep Gnome 1.4, you'll have some very significant problems installing newer apps like Inkscape. On my home system I've given up and upgraded to Mandrake 9.2 and Gnome 2.x for this reason, even though I strongly prefer Gnome 1.4 over 2.x, because I felt the need to be able to run modern apps.) And like I said you can sometimes encounter issues tracking down where Mandrakesoft chooses to put things. The initscripts are less than altogether simple, and stuff like that.
But everything *works*. Well, once you make certain configuration adjustments, like replacing metacity with sawfish, then everything works. With Gentoo I had more trouble getting sundry things set up.
I did learn a lot from using Gentoo, though. I'd recommend anyone with even a moderate interest in peeking under the hood should experiment with Gentoo for at least a month. I'm glad I did. But I find that with Mandrake I spend less time fussing with the configuration and more time getting stuff done, or reading slashdot.
> Last time around, a change of only about 200 voters would have changed > the outcome. (The difference in Florida, IIRC, was 400 votes, and half > that number needs to change.) Out of a hundred million voters, that's > 0.0002%, so you were (amazingly) overestimating the quantity needed.
You're being inconsistent here. If you take the number 400 (thus 200) from the very small area in Florida where the outcome was very close, then you have to calculate the percentate that they constituted based on the number of voters there, not based on the number of voters in the entire nation. If 0.0002% of the voters in the entire nation switched, spread out over the whole nation proportionately, the outcome would have been the same.
And anyway, an election anywhere near that close is unusual, and it's very unlikely we'll see that again this time. Heck, if Kerry doesn't come up with a better theme than "I'm different from Bush", it could be a landslide. That kind of campaign gets you Burger King's market share. Think Dukakis.
Slashdot Mirror
Black paint.
I've found that bringing in alien paradigms can be effective. For example, if
you're writing in C, simulate functional programming mechanisms, such as
closures. This by itself is not enough, because someone could be reading the
code who knows Scheme, but it is nonetheless effective, because it will throw
a significant portion of your audience into unmitigated confusion. Only people
familiar with the alien paradigm will even have a *chance* at understanding
what the code is doing.
> Anyhow, I was perusing the code. It was pretty sloppy, one-letter variable
> names, multiple statements crammed together on single lines.
Done correctly, multiple statements on a line can actually make code easier
to read, by allowing more of the context to fit on the screen at once. It's
harder to defend one-letter variable names, except perhaps for very tightly
scoped lexicals, where you can see every reference to the variable in a small
handful of lines.
> So I get to this one section of code. I can't even remember how it worked
> now, but it was this convoluded for() statement that flipped a flag and
> did some weird ass computation. It took me about 10 minutes of "stepping
> through" it in my head to figure out what it was doing.
Heh. C-style for loops are a pet peave of mine. Fortunately, they're going
the way of the dodo in Perl6, and all for loops will really be foreach loops.
(Oh, you're still programming in C? I'm sorry.)
> It was calculating leap-years. I actually stared at it in shock, imagining
> how much time and energy this kid spent figuring out the worlds most
> assinine way to figure out if it's leapyear. I would have just wrote
> "if (year%4 == 0) { days_in_feb=29; }" or something of the sort.
That would get the answer wrong sometimes. I would write something like
use DateTime; if (DateTime->now->is_leap_year()) { whatever() }
> The fact of the matter is that once you leave your PC unattended, no
> amount security is going to save you.
There are some things you can do, they're just a much bigger pain than locking
up the system in a room with tigers and cobras and security cameras. For
example, you can use an encrypted filesystem that requires a long key to be
entered on the keyboard at boot time. That leaves you open to a sniffing
attack with a KeyGhost or similar device, or clever placement of a hidden
camera to record your typing in the key. There are ways to guard against that
too, by using a challenge and response type method to unlock the encrypted
filesystem, but that can get very complex and require quite a lot out of the
user. (Of course, if you don't reboot very often, that would help minimize
the annoyance factor... but then you don't use the authentication system
very often, so you're very liable to forget it, at which point you now have
to write it down and lock that up in a safe...) This amounts to quite a lot
of hassle, and it really would be simpler to lock the system in a room with
dogs and guards.
Additionally, all of this stuff still allows the user to boot your hardware
(albeit not your software) and probably get onto the network, though, so
then you have to harden your network against internal attacks, which means
careful firewalling and stuff. Which, admittedly, you theoretically should
do anyway, in principle, but I suspect most of us don't.
> the PC in question needs to be set up to boot from USB
Actually, I just noticed that it comes with a bootable mini-CD to get around
that issue. I suspect you'd end up using that most of the time. Does anyone
know whether the CD then boots the USB drive as the root partition, so that
you can function as if you'd booted from it directly, once you're booted? If
it does, then that would give it some advantages over other LiveCD + USB drive
options (e.g. MandrakeMove), since you could e.g. easily install upgrades.
> Perhaps it's a credible alternative to a remote login to your main computer.
m oments thing and becomes rather
> You are still going to need to find an existing computer to plug this thing
> into though.
That wouldn't be a problem in North America (when was the last time you were
more than three rooms away from a computer, really?), except for the fact that
the PC in question needs to be set up to boot from USB, which narrows down
your options rather a lot, since approximately zero percent of the installed
base of PCs out there are set up to do that. It's not even an *option* on
most BIOSes, and one imagines that even on systems with the capability the
default would still be to boot via a more traditional source, such as the
first floppy drive, or a hard drive or ATAPI CD-ROM drive on one of the IDE
channels.
Many modern systems have, in theory, the capability to boot off the network --
but it's seldom the default. You could carry around a smallish LTSP server
and a crossover cable, but plugging it into most systems wouldn't get you
anywhere, because they're not set up to boot that way.
Of course, if you have unobserved physical access to the system you could
*alter* the way it's set up, but at that point this ceases to be a benign
hey-can-I-borrow-your-hardware-for-a-few-
more invasive. And there are still a lot of systems that don't even have
the capability to boot from USB.
It's an interesting concept, but the LiveCD + portable data drive solution
(a la MandrakeMove) seems more realistically useful at this point.
> If the paper is turned in as a printed document, the teacher will have
> to re-type the entire thing and Google each part for potential matches.
You're assuming we need to catch all the students who merely plagiarize small
bits and pieces, _basically_ writing the bulk of their own paper, or at least
putting it together, but without doing proper citations of their sources.
If all we want to do is catch people who copy most or all of the paper pretty
much verbatim, then we don't have to search for each and every part of the
paper; it's enough to look through the paper, pick out a couple of long phrases
that seem like they would be reasonably close to unique (i.e., not something
like "it has been discovered that", which would turn up lots of false hits),
and do phrase searches for them. Any matches that you find, you check to see
if more of the paper matches that source than just that phrase.
That's the good news. The bad news is, Google's only going to find you
mostly stuff that's readily available on the web. If the student copies
a Wikipedia article mostly-verbatim, this will find it. But there's quite
a lot of stuff out there that Google doesn't index. If the student uses
Electric Library or EbscoHost or SIRS to find a journal article and turns
*that* in as their paper, then finding the student's source is not quite
as simple as a single Google search. It can still be done, but it's harder.
> > One last thing though - when (if ever) will Mozilla mail change away from .mbx/mailbox files and move to something like what Sylpheed uses
> > using
> > (1 file per email).
>
> Wouldn't that create a ton of overhead with searches
It would also create a ton of flexibility with searches. (You can e.g. easily
look for messages stored in a given folder within a certain date range that
contain a certain email address in the From: field and also contain a certain
word in the body -- without doing hardly any parsing.) However, Mozilla Mail
has never really been about flexibility; it's more about simplicity. People
who want flexibility spend the three weeks of initial learning overhead so
they can use Gnus, at which point they can do pretty much anything -- and even
more than that if they know some lisp. Mozilla Mail is for people who just
want to start reading mail right away, instead of reading a lot of info pages.
(Yeah, I use Gnus. But I use a mozilla.org browser, and I'm experimenting
with Sunbird.)
> I think they need to use different animals for each program though.
> Firefox, thunderBird, Why not something like aquaMoose?
SunMarmot? Too dorky. SunShrew? Too grouchy. SunTzu? Too obscure, and
potentially offensive. SunWoodchuck? Too long. (Just SunChuck perhaps?)
SunDeer? Closer, but I don't like a product name ending in r; it's too easy
to get sloppy on the pronunciation. SunShark? Too toothy. SunGorilla?
Overpowered. SunSquirrel? Nuts. SunSwift? But that's a bird. SunMouse?
Too timid. SunDog? Too ghetto. SunSloth? Too lazy, especially for a
calendar app -- people would be afraid of missing appointments. SunCat?
Too many people don't like cats. SunTiger? We're not Apple. SunLynx?
There's already a browser called Lynx. And so on.
There are some possibilities, though. Personally, for a calendar, fitting
in with the Firefox/Thunderbird theme, I like Stormcrow, because it warns
you of things. Then again, if it were me, phoenix/firebird/firefox would
be named something more original, like FlameThrush, and Thunderbird would
be named more originally too, like BoomSwift (notice, another kind of bird),
and then the calendar would have different lines to work along and could be
something like DayOwl or FlashMartin or, indeed, StormCrow.
> Linux is NOT A DESKTOP OS
Linux isn't an OS at all; it's a kernel. Nevertheless, it was developed first
and foremost for desktop use (albeit, for exteme geek desktop use). It had to
develop for several months before it approached server usability, and for quite
a while before it was really considered production-quality for servers. Even
today, it is arguably not as well suited for servers as some other OSes (most
particularly BSD), but it's a great desktop OS (for geeks -- though the
extremeness of geekhood involved has declined substantially; it's no longer
even really necessary to know C).
> My user agent string: "All your base are belong to us."
I prefer, "I don't have a browser. I type TCP packets directly on my keyboard, you insensitive clod!"
It's because *nobody* can stand Real. Their software has more nag screens
than anything else you can download, and, additionally, it's feature-poor,
doesn't work well, and uses (surprise) its own proprietary format in an
attempt to lock people in. If a decent company with a useable product had
reverse-engineered Apple's DRM system, public opinion might have been rather
different, but Apple has a good reputation, and Real has such an extremely,
utterly bad reputation, they make Microsoft look like a the poster boy for
popular companies.
> Usenet isn't just any old NNTP connection, it's a public, widely used
> network of NNTP and UUCP and whatever else connections
I'll go for usenet being a proper noun, but it's lowercased anyway because it
derives etymologically from Unix culture and so is case-sensitive.
> The Internet is the same- it's not just a bunch of machines connected
> together via TCP/IP, it's the collection of machines connected together
> via TCP/IP.
The internet we have happens to be the only worldwide internetwork, but it
would certainly be possible to have another. Imagine for a moment that during
the third patent war in 2137, all interoperability between open-source systems
and commercial systems is lost at the IPv8 level. Voila, two internets. In
the more immediate present, imagine an internet of IPv6 hosts, and an internet
of IPv4 hosts. Hosts that support both protocols would be on both internets.
Or, if you're into sci-fi, imagine aliens that have their own internet, based
on their own technology and protocols, and negotiations (and, if you prefer
to imagine it, war) over how to get the two internets joined.
There's nothing inherently unique or singular about the concept of an internet;
we call it _the_ internet merely because we only have one, just as in a small
community you find people speaking of the library, the lumber yard, and so on.
I have not yet lived in a community small enough to speak of "the restaurant".
> a LAN is a net, but not the Net
A LAN is a network, but it is not the only network -- but in a given place
it might be the only one under consideration; at work, for example, we only
have one network, and I regularly speak of "the network" at work, when I
mean, "our LAN". It isn't whether there's only one or not that makes a
thing a proper noun. If that were the case, universe would be a proper
noun (since, by definition (everything that exists), there can only be one)
but it's not, because the term refers to a concept (i.e., is a common noun),
even though in practice there turns out to be just one of them. On the other
hand, there might be *lots* of people named Bob Smith, but it's still a
proper noun (or a proper noun phrase).
What next, will they also stop capitalizing "usenet" and "web browser" and
"service provider" and "post office" and "bank" and "grep" and "sort"?
If internet were a proper noun, there would be other worldwide communications
networks with other names. (No, don't say the phone network; phones and
web browsers transmit their data over the *same* network, they just do it
rather differently.) TCP/IP over avian carriers never really caught on,
for some strange reason, so we only have one internet. But if we had did
have two distinct internetworks, they'd both be internets.
usenet is an edge case -- it really probably ought to be a proper noun, but
it comes out of Unix culture, so it's lowercase as a matter of case-sensitive
spelling, even at the beginning of a sentence, like grep and sort (when sort
is used as a proper noun -- when sort is used as a verb or a common noun it's
from standard English and is capitalized according to the normal rules).
And yes, this is consistent with the normal rules of English, in the sense
that the normal rules of English allow for exceptions based on sepcial rules
pertaining to a given etymological source. There are also many English
words that are CamelCased -- and I don't just mean computer words, either --
because of their etymology or the particulars of the field they come from.
Similarly, words derived from foreign languages often form their plurals
specially or are pronounced specially according to the rules of the source
language or field, e.g., pianos and filet, respectively. case-sensitive
spellings from Unix culture are consistent with this.
> Our faculty library still uses vt220's.
We've got VT510s here for our library catalog, and they work great for that.
I'm not sure the public would accept them as an internet accesss solution,
though, much less for word processing.
> I've never known a library to use 19" monitors for anything
We are moving in that direction. 19" CRTs (18" viewable) or 17" LCD, depending
on physical space requirements. We quit buying 15" CRTs several years ago and
are now choosing 19" over 17" a significant proportion of the time.
A 19" CRT these days barely costs any more than a 17" CRT, and they are
significantly easier on the eyes of many of our patrons, who do not all have
the eyesight they used to have. (Seems to be a bigger problem than when I
was a kid. I blame it on television; don't know whether there's actually
a cause-effect relationship there, but it feels good to blame television for
something more concrete than moral degradation.) We find that going to 19"
allows us to set the resolution to 1024x768 without making most folks squint;
on a 17" monitor, 800x600 is pushing it for many and 1024x768 is right out;
on a 15" monitor (of which we no longer have any except all-in-one units such
as iMacs[1]) the resolution really can't be set any higher than 640x480,
which isn't even an option on newer versions of certain OSes.
LCDs would be slightly better due to the extra inch you get out of their more
honest screen size notation, except that they look like exrement at normal
resolutions, basically forcing you to crank them up to their "native" res,
which results in a lot of squinting. But we nevertheless use them sometimes,
particularly in cramped locations where a 19" CRT would be unweildy. I look
for LCDs with a relatively low ratio of native res to diagonal measure, to
minimize squinting. (This is counterintuitive, since looking for smaller res
numbers violates the "more is better" principle. I guess in this case the
normally-inane "less is more" principle actually applies. Phooey.)
[1] And VT510s, which are even *smaller*, but those are amber-on-black, which
significantly cuts down on eyestrain right there, and in any case we'll
be moving away from them circa 2005-2006 when we migrate from Galaxy to
a more modern ILS.
> That'll be a couple of SSH terminal sessions and some FTPing then
Actually, I'm thinking one Cygwin window containing an X session, which you
use with X11 forwarding to display the apps running on your *nix system.
Oh, wait, you could just use *nix then. Remind me what I want XPSE for?
Oh, yeah, because I wouldn't want people thinking I'm a computer geek who
uses a weird custom OS that didn't come with the computer, now would I?
You will not believe how bad The Creeping Terror is until you watch it yourself.
I cannot describe its badness (not that that's going to stop me from trying...)
It makes the special effects in The Blob look stunning by comparison. It makes
the acting in The Crawling Hand look like Academy Award material. It makes the
character development in War of the Worlds look deep and meaningful. It makes
the plotline of Spy Kids 2 seem almost plausible. It makes the camera work in
Return of the Killer Tomatoes look impeccably professional. It is by far and
away the most amazingly bad movie ever made. From the concept to the
implementation, from the writing to the scenery, from the acting to the
directing, it had *better* be the most impressively pathetic movie, because
if there is a worse one, I think I will wet my pants when I find out about it.
In a nutshell, The Creeping Terror is about these alien things that come out
of this landed spaceship. They look remarkably like people under a blanket
crawling around, only much slower. Occasionally you can see shoes under the
edges. Whenever these creeping things encounter anyone, the encountered party
screams and freezes, because if they ran (or walked, or indeed crawled) away,
the movie wouldn't have a plot. So they freeze, and scream, and wait for
the terrible creeping blanket to come right up to them, at which point they
have to crawl up into the thing to be eaten. This happens a whole bunch of
times, while scientists and police and whatnot poke around the spaceship in
a completely vain attempt to figure out what's going on. Eventually they
find the first one (which got out of the spaceship when it landed) and figure
out how to disable it, and then they inadvertently unleash the second one
(which was up to then restrained in the ship), and go through the whole big
thing *again*, and then when they disable the second one, the lead scientist
realizes that it sent a signal back to the ship -- so they go back to the
ship and try to shut it down before it transmits, but they fail, so it
transmits (something) into space, causing the lead scientist to conclude
that someday more of these things will come, and we'd better be ready, and
then, mercifully, the movie ends.
And if you think that sounds long and boring and lame, that's the executive
summary. Remember how I said the movie is so bad, I cannot describe its
badness? Yeah, I meant that. I've barely scratched the surface. Don't
believe me? You must watch the movie. It will exceed all your expectations
of badness and leave you stunned, in denial that any movie could be so bad.
According to my analysis, only about 50% of spam is written in a Latin-derived
character set[1], and that includes all that Spanish spam from Central and
South America, all the money scams written in mostly ALL CAPS that are
generally believed to come from Africa, and anything that comes from Western
Europe or Australia (not much from there I think), and any spam written in
English coming from Asia, which is doubtless no small amount, in *addition*
to everything from the US.
*Most* of the other 50% is written in decidedly Asian character sets (though
there's a little Cyrillic spam and the occasional Greek message). As much as
25% of all spam is in the GB2312 character set alone. I have a really hard
time believing that's coming mostly from the US.
Maybe my sample is skewed (admittedly, I'm looking mostly at spam sent to
one address), but it's a pretty good-sized sample (1.3GB).
[1] For UTF-8 messages, I base my "character set" data on the actual
characters used, which in spam are usually ideographic or syllabic
characters from Asia; most English-language spam is written in ASCII
or Latin-1 or something similar. Most English-language UTF-8 messages
are from mailing lists related to the open-source community. I suppose
this is probably because US localized builds of Outlook and its ilk
don't send unicode by default.
> Most likely it will come with something new and fancy called Windows
Yeah, I used to use that. But even then, I usually kept Emacs filling the
whole width of the screen. I find that while it's useful to position two
or three or even four windows vertically above one another, horizontal
tiling is of less utility than the ability to work with wider content.
> Or you could take 20 mins to figure out the security settings you need on
> an XP box to stop the user from installing thier own programs.
Right, so then you'd only have to worry about the apps that come with Windows,
such as MSIE and MSOE. Fortunately those apps are very secure, as we all know,
so the whole solution would be very robust... or something.
> I'm pretty sure this disqualifies me from being a philosopher
No, you can be a philosopher without drinking alcohol if you have something
else you drink instead -- say, coffee or tea -- or some other quirky behavior
that causes other people to write you off as eccentric. (This is a key
difference between a philosopher and a thinker of some other sort, such as
an analyst or a strategist: people can respect an analyst or strategist,
because they believe his thinking has value; they look askance at a
philosopher, because they believe his thinking is off the wall.)
> Is this the first OS to go beyond version 10?
:-)
In a word, no.
> I know there is MacOS X, but that's kind of randomly chosen name (more
> than a version number), no?
Lesse, there was System 6, System 7, MacOS 8, MacOS 9, then Mac OS X, which
was at first 10.0 then 10.1 then 10.2 then 10.3 and soon 10.4. Looks like a
number to me.
Nor was Mac the first OS to hit double-digit version numbers, and I think
Solaris was there before Mandrake. Heck, Emacs is at version 21 now
I did try Gentoo, and the experience was informative, but ultimately I came
back to Mandrake. It's slightly more of a pain with Mandrake to install
bleeding edge stuff (except for major apps like Mozilla, which are both easy
and straightforward), and you spend slightly more time hunting down where
the distribution puts things, but I found that for me, Gentoo outweighed its
advantages in those areas with some disadvantages of its own.
Foremost, things just aren't as well integrated with Gentoo, in terms of the
way things are set up. I think this mostly has to do with configuration --
sure,you have to reconfigure things with any distro to match your preferences,
but with distros like Mandrake, the initial state is that the various
components are all set up to interact well with one another. With Gentoo,
I had some trouble getting things to play nicely with one another.
I expected Gentoo to solve dependency hell for me; it doesn't. In more than
one case I had to manually track down packages that were causing emerge to
fail, and individually install (or uninstall!) them. In one case I had to
hunt all over the internet and eventually find that in order to emerge world
I had to first uninstall certain core Perl modules. That should never be.
This is different from the situation in Mandrake, but IMO it's not better.
It was also not always obvious what package I had to install to get a certain
thing. I had an horrific time figuring out how to install the themes for
Gnome. With Mandrake, this is not a problem.
Mandrake has its problems, sure. It's not the most secure distro, and
there can be problems if you only want to upgrade some components and not
others. (For example, if you want to keep Gnome 1.4, you'll have some
very significant problems installing newer apps like Inkscape. On my home
system I've given up and upgraded to Mandrake 9.2 and Gnome 2.x for this
reason, even though I strongly prefer Gnome 1.4 over 2.x, because I felt
the need to be able to run modern apps.) And like I said you can sometimes
encounter issues tracking down where Mandrakesoft chooses to put things.
The initscripts are less than altogether simple, and stuff like that.
But everything *works*. Well, once you make certain configuration adjustments,
like replacing metacity with sawfish, then everything works. With Gentoo I
had more trouble getting sundry things set up.
I did learn a lot from using Gentoo, though. I'd recommend anyone with even
a moderate interest in peeking under the hood should experiment with Gentoo
for at least a month. I'm glad I did. But I find that with Mandrake I spend
less time fussing with the configuration and more time getting stuff done, or
reading slashdot.
> Last time around, a change of only about 200 voters would have changed
> the outcome. (The difference in Florida, IIRC, was 400 votes, and half
> that number needs to change.) Out of a hundred million voters, that's
> 0.0002%, so you were (amazingly) overestimating the quantity needed.
You're being inconsistent here. If you take the number 400 (thus 200) from
the very small area in Florida where the outcome was very close, then you
have to calculate the percentate that they constituted based on the number
of voters there, not based on the number of voters in the entire nation.
If 0.0002% of the voters in the entire nation switched, spread out over the
whole nation proportionately, the outcome would have been the same.
And anyway, an election anywhere near that close is unusual, and it's very
unlikely we'll see that again this time. Heck, if Kerry doesn't come up with
a better theme than "I'm different from Bush", it could be a landslide. That
kind of campaign gets you Burger King's market share. Think Dukakis.