Yes. Where I work (a small public library), the word "upgrade" means "replace with a newer system", and sometimes the old system then gets moved to a different location and used for a different purpose. We have the following:
1 Windows 3.11 for Workgroups (to be retired "soon")
2 Mac System 7 (yes, really)
3 Windows 95 (I think these are all OSR2)
2 MacOS 8
4 Windows 98 (I think at least three of these are SE)
3 MacOS 9
1 MacOS X (10.1.5; this is the director's workstation)
1 Windows XP / Mandrake 8.1 dual boot (this is my
workstation; I use Linux mostly but keep XP around
for testing purposes, and so I can better understand
users (patrons) who have that OS.)
1 OpenVMS 7.2-1 (this is our main catalog system)
1 Linux 2.2 (SMC TurboLinux; our low-demand cgi server.
Our main web server is hosted by our ISP, using
WebStar I believe.)
There is also a Windows PC maintained by UAN, which I _think_ is Win98, but the library doesn't have anything to do with upgrading/purchasing/replacing that one; it belongs to the state's accounting network. It runs custom software that was apparently written for DOS, and it hangs capriciously from time to time. The only thing I want to know is why it has an APC UPS, when the software wedges WAY more often than the power goes out.
I would be curious to know whether fragmentation is more extreme in small or large organisations, and whether there is a correlation with what the organisation does (i.e., what the company does for its money or what the government or nonprofit organisation does to justify its existence).
Don't get me wrong; I _love_ Perl. But the way Perl 5 does OO feels tacked on at best and really seems quite tortured. IMO, they should have waited for the major syntax breaking in Perl 6, which will (hopefully) allow them to do OO in a way that can be more like a core part of the language and less like a really strange syntax for hashes.
I certainly agree that Perl has improved since version 2, or even since version 4, and that more improvements can potentially be useful. If I understand correctly, early versions of Perl 5 didn't have very good language support... that surely seems worth fixing, and I'm sure there are other things. I'd *really* like to see the addition of textbuffers (a la elisp), which, combined with dynamic scoping (which Perl already has), can be extremely powerful. (Not that this is in the plans...)
Perl 6 is going to be such a huge change, it has me both excited and scared at the same time.
Funny you should mention that. I've been saying for some time that mithril is actually just the old name for titanium steel, and that adamantium would be something strong but more flexible, like an alloy that includes organic matter (plasteel, anyone?), but maybe this liquid metal stuff could be it.
> I've run Cat5, I have a couple switches and an old hub. > I know that Cat5e and Cat6 are better than Cat 5. But > that's about all I know
As a consumer, you don't _need_ the technical details (unless you're just curious, in which case I'll let you do your own web search). One thing you do need to realise is that any given network segment is only going to be as fast as its slowest link. A switch can make each connected node into effectively a separate segment (so that one slow workstation won't slow the whole network), but a non-switching hub won't.
So, if you have a 100BaseT switching hub, that's 100Mbps, and no cabling you can put in will make it go faster. (You can make it go _slower_ by using less-than-cat-5 cable, however.) If you've got a faster switch and faster cable, you still need a faster NIC to go with it, for each system that needs to take advantage of the speed increase.
For Perl, I want the curley braces and dollar sign in three of the "easy" positions (where letters normally go). Slash and backslash are negotiable.;-)
Dvorak aside, is the difference between this layout and QWERTY significant enough to be worth switching? (Most people, remember, would be switching from QWERTY, not from Dvorak.)
Personally, I doubt it, but not for the reasons you give. In particular: with QWERTY, I can type over 50 wpm if I'm reading it off, and it's fairly standard English (as opposed to, say, Perl). But most of the time I don't type 50 wpm. Most of the time, I spend time thinking of how exactly I'm going to word things. I'm not talking about sitting idle between clauses, but I type more slowly because I'm thinking as I go. Typically I get somewhat less than the 50 wpm I can physically get. (Offhand, I would guess that I tend to get 20-35 wpm, depending on the situation, and in special cases less. Of course, sometimes I do sit totally idle for several seconds, but I'm excluding those times from the average.)
For this reason, I'm not terribly interested in going to great lengths to optimise my key layout for sheer speed. I'm a *lot* more interested in optimising it for comfort. In particular, reducing the amount of strain on certain fingers that with a normal layout frequently have to be hyperextended. The layout I'm using at the moment has shift and ctrl on home positions; since I adopted it, my left pinky doesn't hurt all the time anymore. At first there was some slowdown, and a few mistakes, but I'm beginning to get past that. I can (under ideal, read-it-off conditions) get 50+wpm with my new layout now. I'm even starting to be able to switch back and forth between my layout at home and bog-standard QWERTY at work, with a minimal adjustment period (a couple of minutes, typically). (It may help that I left _most_ of the keys alone, and just moved certain strategic ones.) Having shift under a home position is to me like having pants that fit; I'm not mostly worried about whether I can walk faster.
> (i.e. have a 'Help' key, rather than "Press F1 for help")
The VT510s where I work (a public library) are set up this way. The catalog-only terminals (for the public) have keys like "Search" (which is really just PF1 or PF2 or something like that, but the software and the keytops are designed for one another). The circulation terminals have keys like "Patron Status" (which is really just F4 or F5 or something, but the software and the keytops are designed for one another). Very nice. Combined with software that quite literally tells you which buttons to push, all but the most obstinate and thorough technophobes have little trouble figuring out how to work the things.
The only trouble with this is, it only works so long as your software and your keytops are made for eachother. The minute you want to use the same computer for thirty different applications, this will never work. Sure, you can have certain conventions. We already have keys reserved for page up and page down, beginning and end of line, and such. Reserving F1 for help and labelling it as such seems reasonable, since most apps follow that already anyway. (The key should probably still say "F1" in smaller print, in addition to "Help" more prominently, in case of deviant apps.) But going much beyond that...
Hey, what about designing a keyboard with an extra row of keys for things like Copy, Paste, and Save, and having them generate Ctrl-C, Ctrl-V, Ctrl-S, and so on. Sure, it wouldn't jive with Emacs[1]... But for most end-user type applications... could be handy.
[1] NOTE: I am not dissing Emacs. (If you weren't
about to reply to what I said about Emacs, you
don't need to read the rest of this.) I would
die without Emacs, or suffer pain at any rate.
I refuse to use any computer without Emacs
installed. I use it to read my mail, as well as
usenet, and have been known to use it to browse
the web. I have written my own (fairly simple)
major mode on more than one occasion. A.tar.gz
of just some of my sitelisp is 161222 bytes. So
don't even start. I've compiled my own Emacs on
at least three different operating systems, in
some cases using not-yet-officially-released alpha
versions of the source, because I _wanted_ the
promised new features. I did the proof of concept
for the color-theme module. In short, I _love_
Emacs. I'm only using it as an example of something
that doesn't use the same conventions as other
applications. I'm fully aware that there are very
good reasons why it doesn't, and that people who use
Emacs don't really need other applications. And yes,
of *course* I know about remapping the keys (and have
done a lot of it, not just globally but also in mode
hooks). But I was referring to the out-of-the-box
state (not that I ever leave anything in the OOTB
state... but some people do, I guess).
> A physically-remappable keyboard could be a good thing. > I'd probably buy one.
I have one. An Avant Stellar. Produced by Creative Vision Technologies. It costs money, but it's a good keyboard.
Now, the act of rearranging the keycaps doesn't cause the layout to be remapped. But the layout can be remapped hardwarily (or with Win9x-based software, provided), and the (same-sized) keycaps can be swapped around. An extra set of keytops are included for the left Ctrl and CAPS LOCK keys, in case you want to swap them even though they aren't the same size.
However, I wanted to swap some other stuff (put shift under a home position...) so a few of my keys don't say what they are. I could remedy that with white-out and ink, if I cared...
> "Please refrain" is somewhat of a dangling clause.
I elided the word "then" from its position as the subordinator for the subjunctive clause. This is standard practice in all dialects of English of which I am aware.
> an inanimate object
My bad. (I've been fooling around with a keyboard layout that is not strictly QWERTY, and have been mixing up i and a lately; it'll clear up as I grow accustomed to the layout.)
> Second: "or whether it had an unlimited capacity" > is awkward phrasing.
Not one quarter so awkward as "whether... or that".
> Second: Ah, that's enough. I think the point is made.
My original point was that the original rant about "moot" was stupid, and that we had better things to discuss than grammar. Apparently I was not entirely clear.
Yes, the word is "moot", but if you are going to correct grammar, and cannot manage to use correct grammar when doing so, please refrain.
> A "mute topic" is a topic that doesn't speak.
Perhaps it would be, or perhaps it would be a topic about which no one speaks.
> I had a partner that used to say that ALL THE > [...] TIME [...]
No, you had a parter who used to say that frequently. I doubt very much whether your partner was in inanimate object, or whether it had an unlimited capacity to continue speaking day and night.
> Sheesh, are people that ignorant and retarded???
Many people are indeed that ignorant; whether they are retarded is a separate (and irrelevant) matter. On a related note, it is incorrect to treble the interogatory mark; one piece of end punctuation, together with the introductory interjection, is quite sufficient to indicate the mood of the sentence.
> Considering that I never actually open many of > the buggers, I am actually quite happy that I > don't have to download their effing images over > my modem connection.
But that's true regardless of your mail client's behavior. And unless you've already _had_ the kind of privacy leak that autoloading of remote content enables, there's no way for the spammer to know what mail client you use, so he's going to spam you regardless.
The real solution to _this_ problem is pre-retrieval filtering. It is possible for a POP3 client to download just the headers of the messages on the server and then decide based on those which messages to retrieve in full. Most mail clients don't do this on the grounds that it means retrieving the headers twice for the messages that are not excluded, but as the volume of spam increases to more than half of all mail, it may be well worth doing. The messages that are excluded from full retrieval by the filter could, if desired, be left on the server for the time being and the headers kept someplace for the user to examine at his leisure and mark for either full retrieval or for deletion from the server. If the user needs to see the body to decide for sure whether it's spam (about 0.05% of all spam plus or minus two orders of magnitude, according to Flagrantly Estimated Statistics), then of course he has to mark it for retrieval, but that will be a minority case.
> If they sent all their crap inline
Plenty of them do that as well, but in that case there's no preventable privacy issue. Spam is still bad, of course.
> Just because a particular technology is misused by idiots, > why should we all have to do without it?
Huh? What "technology" are you talking about doing without? The ability for your mail client to go running off to a message-specified remote server without user action? I suppose you also don't want to do without the "technology" that lets your mail client execute binary attachments without user action?
Obviously, if the message is legit, the user can always hit the "load remote content" button, open the thing in a web browser, or whatever.
Though I prefer just sending a URL and letting the user doubleclick it to launch his web browser of choice, or if his mail client doesn't support that he can copy and paste. Spammers can do this too, but unless the user takes action to enable it, there's no privacy leak *and* less bandwidth consumption.
If they've replaced Linuxconf with something else, that's almost certainly a good thing. All my experiences with Linuxconf, across three distributions, using the text-mode version, the Gnome version, and the web interface, have been bad.
Not only does it leave out some of the things users often want to configure, and obfuscate others, but it also (as of last time I used it, which admittedly is not recently) hangs and dumps core way too often for comfort, to say nothing about creating inconsistent state so that the easiest way to recover is to reinstall.
What is needed is a tool that integrates the Gnome or KDE (whichever session the user is in) control panel together with graphical interfaces for enabling and disabling services (in a way that not only starts/stops them but also changes the relevant startup scripts so that their started/stopped-ness persists after reboot), GUI tools for setting up multibutton mice and scroll mice, changing resolution, colour depth, and refresh rate of the X server, and so on. Anything you can do from the Windoze control panel should be included.
I like the tree-on-the-left panel-on-the-right approach of the Gnome control panel, but besides Gnome, the rest of the system config needs to be gathered there too -- at least, anything a newbie might want to configure.
Dvorak makes some of the same mistakes as QWERTY, overusing the pinkies and underusing the thumbs, among other things. In particular:
Certain buckies (shift especially, and to a lesser extent control) are used more frequently than the lesser-used letters, but they are stuck out of the way where pinkies have to be hyperextended to reach them. This is *bad*. These keys need to be easy to reach, _especially_ since they are often held through several consecutive keystrokes.
Only one thumb is used profitably. It would be really nice to split that spacebar in half and use the other half of it for something... For something you often want to hit at the same time as other keys, such as shift, this would be a huge improvement. In fact, if the keyboard is split in two (as it should be, really), there's no particular reason the left thumb can't have a choice of two keys to hit, which could be shift and perhaps ctrl.
Some of the seldom-used punctuation (square brackets and curly braces, for example, which seem to pop up mainly in Perl) are easier to reach than more often used punctuation (such as backslash/pipe, which are used constantly on some systems).
The Windows key on 104-key keyboards is too easy to hit accidentally, when reaching for another key. But hitting it disrupts whatever you are doing. It should be up in the extreme top row, with things like PrtSc and Pause. (Besides Windows, this also effects Gnome, at least with some distributions.)
The extra keys between the main keyboard area and the keypad are in the way and should be removed, or at least moved to the outer edge beyond the keypad. They duplicate funtionality already extant in the keypad since 1980 if not before, in a more difficult to use layout. (On the keypad, you can easily reach all the cursor movement keys without moving your hand. In the area to the left this is not the case.) Their only purpose is to increase the distance the hand must travel to get from the home position on the main area (for data entry) to the home position on the keypad (for navigation).
I'm using a modified keyboard layout, described in another post, and it's some improvment, but I had to work with the same _physical_ layout as everyone else, and could have done much better given the opportunity to physically redesign the keyboard and, for example, make better use of the second thumb.
> What? You mean getting a key to generate a different > character from the one that's printed on it?
The keytops on Avant keyboards are interchangeable (well, ones that are the same size are interchangeable). That doesn't get you a pounds symbol, but you can always use white-out and ink to make your own keytop display. Though if you often look at the keys while typing, you may not be enough of a keyboardist to fully appreciate the merits of a real quality keyboard.
The _big_ advantage of Avant keyboards is the full remappability and programmability. I've got an Avant Stellar, and I've got it remapped to reduce strain on my left pinky, by repositioning the frequently-used shift and control keys under my left and right pinkies, respectively. The semicolon/colon (which is normally where I now have ctrl) I moved up to the left square bracket position, and I used the right square bracket position for backslash/pipe, and moved the square brackets down to bottom-row positions, since they are less frequently used. The a key (normally where I now have shift) I moved to where k normally is between j and l, and the less-frequently-used k I put where left shift used to be. I repositioned the left window key to the top function-key row, where I can find it easily but won't hit it by mistake, turned the right-shift (which I never use as such) into an additional backslash/pipe key, turned the capslock (which I never use on purpose) into an additional control key. The resulting layout takes some getting used to, but the frequency with which I have to take my pinky out of the home row is *vastly* reduced, which makes for more comfort over the long term.
Of course, you may have your own set of peeves about the standard keyboard layout. The point is, with an Avant keyboard, you can remedy them. Oh, it does macros, too, and with the Stellar model you get an extra set of function keys, ripe for macroing.
One caveat: most remapping can be done with no special software, using the hardwired key sequences, but if you want to remap right ctrl, left shift, or 1/!, you need to use the included remapping software, which is Win9x-based. I have a copy of Win95 on another partition, that I was able to use for just such an emergency; this means that I have to reboot (before and after) each time I want to make changes to my layout, but once I got my layout worked out I don't anticipate frequent changes.
This full remappability/programmability is what makes CVT's keyboards special. Yes, they're also quite good quality physically; there are reviews out there on the web, so I won't duplicate them here. But the versatility is the killer feature. I was *happy* to pay more than the usual keyboard price for this model. CVT is for keyboards what Lian Li is for cases. If you don't think keyboards are a big deal, if you're satisfied with your $40 ergonomic keyboard (or even with a $9 cheapie), then an Avant may not be for you. If you find yourself dissatisfied with the shortcomings of traditional keyboard layouts, disgusted with the quality of membrane keyboards that succumb to humidity and have to be replaced frequently, in general, if you just want a better keyboard, I see an Avant model in your future. (I'm just a satisfied customer; I have no affiliation with the company.)
The test: if you are looking at a webpage, and need to scroll, do you
A) Reach for the mouse.
B) Use the mouse without needing to reach, because your
hand was on it already.
C) Keep your hands on the keyboard and use the relevant
keys to scroll, because they're closer and easier to
reach than the mouse. If you answered C, you will appreciate a quality keyboard. If you answered B, you won't care. If you answered A, you are a borderline case, and may or may not find a quality keyboard worthwhile, depending on your budget.
Not "only for beginners" per se, but Mandrake is especially well-suited for users crossing over from other OSes, especially from Windows. This doesn't mean it can't be used for other things, only that the places where it shines out and is notably better than other distros are mostly to do with being friendly to the unwashed masses. Yes, it's also optimized for Pentium... but in that regard it can't claim to be better than Gentoo, for example; Mandrake is, however, more friendly to newbies from Windozeland than Gentoo. That's the most significant strength of Mandrake (and it is an important thing).
Second-tier is not a _bad_ thing. But yes, I'd say Suse is a second-tier distro. (So is Mandrake, as far as that goes.)
Overall, there are probably only two first-tier Linux distros (Debian and RH), as I estimate it. What makes them first-tier is that they form the foundation for other distributions that are based on them. So the second-tier and third-tier distros fall into three categories: those that are based on Debian, those that are based on RH, and those that go their own way. (These can be further subdivided, of course.)
I'm not a Suse user, but my understanding is that they go their own way, not tracking closely with either RH or Debian. Which is fine. But unlike RH or Debian, I don't know of any distros that are based on Suse.
United Linux seems to me to be several second-tier distros getting together to form a third first-tier distribution. Now there will be numerous distros that can be classified as UL-based, as well as those that are RH-based and those that are Debian-based, and of course those that go their own way (e.g. Gentoo -- and speaking of Gentoo, I ran across the word "gentoo" in a book yesterday (_Endurance: Shackleton's Incredible Voyage_), wherein it seems to be a type of penguin.)
I don't have a problem with that. As the popularity of Unix in general and Linux in particular continues to grow (as has been happening nonstop since the original development of Unix at Bell Labs), there will be room for more first-tier distributions, as well as more second-tier distributions, more third-tier distros (yes, the line between second-tier and third-tier is probably pretty blurry), and more specialty distributions as well, to say nothing of more bundled-with-hardware distributions.
For now there are few enough systems sold with Linux-based distros preinstalled that they use existing distros; when it really catches on, the hardware vendors will put their own distros together, just like the bizzarroid bundling they've been doing all along with other kernels (both unices and other).
Like with all unices, they'll have more freedom (than with, say, MS OSes) to make the branding really pervasive, with their corporate logo not just on the default wallpaper, but built into the web browser (which has its start page set to their corporate site), embedded in the panel background, all over the GUI administrative apps (think: "Compaq Control Center"), all over the bundled office suite, the bundled email client, the bundled music/video player, and so on and so forth. Think of the Gnome foot or the KDE gears being replaced with the hardware vendor's logo. Think of the default signature in the bundled email client advertising the vendor, unless and until the user changes it.
There are huge opportunities for hardware vendors here. When the software is a commodity, you can brand it any way you want, and you don't have to share your branding with another company. Sun and Apple have been doing this for a while, with more expensive hardware, but there's no reason it can't be done with low-end hardware and by other vendors.
> Most spam I get these days is HTML phone-home style. As soon > as I read it, it's off requesting images from somewhere.
If your mail client lets it do this, your mail client is br0ken. Good mail clients NEVER retrieve content from elsewhere on the internet during the course of merely reading a message. That's a fundamental privacy flaw, a serious performance issue, and in certain cases even a security hole (though not with mere images). There is also no legitimite reason for it in a normal message. (If you *must* have inline images (a highly dubious point; attachments will do for way more than 99% of cases), that's what the disposition/inline MIME thingy is for.)
> > For myself, I'm sticking with my existing dialup > > account until I can get true broadband, something > > better than cable modem (such as DSL). > > What a ridiculous conclusion.
I did say "for myself". Also note that I'm considering the costs of switching service providers (e.g., tracking down hundreds of people and getting them all my new email address, tracking down all the search engines with my webspace indexed and getting them to index the space at the new provider, and so on), and when all of that is thrown in, cable modem service (at least in this area) is just not good enough to warrant switching, for me.
Also, several repliers have cited impressive download speeds for cable modem service, but if I went broadband I'd want bandwidth in the other direction, too...
> When you are talking viewable, do you mean in reference > to the laptop (ie LCD) world, or the desktop (CRT still, > for the most part) world? I am confident you know the > difference.
I understand the difference. I said viewable, because my current monitor is 18 inch viewable (a "19 inch" CRT), and if I got a laptop for my next system, I'd prefer not to lose screen real estate. That's why I chose 18, instead of 17 or 19, as the value for a future laptop to match in order to be an acceptable replacement for my current (desktop) system.
> If Linux is such a priority,
Not so much Linux specifically as Unix in general, but I said Linux because it tends to have better hardware driver support than (say) BSD, particularly on low-end consumer-type hardware.
> and Linux is still quite a bit of command line, or > maybe you write a decent amount of code, than I am > not sure I understand you wants/desires. Gaming?
I'm not sure I understand the question. I was talking about being able to get all the hardware (NIC, sound if possible, and so on) working with the a Linux kernel. Hardware that doesn't work except with Windoze or NT is of substantially decreased value to me.
> And your gripe about poor viewing angles has been > a moot point for at least a year or two now.
Such has not been my experience. Current LCD screens (at least, the low-end ones) do not have viewing angles that impress me much. In particular, if you move your head as little as thirty degrees (relative to the screen) it changes the colour ballance. It's not as horribly bad as it once was, but LCDs still need some improvement to catch up to CRTs.
Still, I'm not a graphics artist, and so to me, screen _size_ is more important than colour accuracy, so I might be able to settle for a laptop screen if it's large enough.
15.7" is really not large enough. I could use maybe use it on trips, but I can't afford to buy both a laptop (for trips) _and_ a new desktop for regular use. Plus, it would drive me out of my mind not having my regular (heavily customised) environment on the laptop. So if I'm going to get a laptop, I'd really want one with a display large enough that I could also use it as my regular system. So, 18" or thereabouts.
I suspect it does. Actually, DSL has the normal rules; it's cable modem service that has different rules.
There are good reasons for this, and it comes down to design. The cable network was _designed_ to deliver lots of the SAME content to every subscriber. Yeah, it can carry 50+ channels, but it takes the same 50+ channels to everybody. Need to add another thousand users? No problem: just tee off the line that's already there. As cable TV grew, they've been teeing the lines for years.
As a result, the lines can handle pretty substantial bandwidth to an individual node, but the _total_ amount of bandwidth on a given subnet has some pretty stiff limits on it. In particular, the total amount of available bandwidth, that has to be split between all nodes, is not necessarily greater than what one node is capable of receiving (unless they place some artificial cap on the latter).
This is backwards from almost every other network. If you have, say, a T1 line, your provider (typically) won't whine much if you use its maximum capacity, because the T1 line itself is the chokepoint. At the other end is a router that's probably on your ISP's core LAN (for small ISPs) or core WAN segment for your area. The bandwidth there is somewhat more plentiful (though not unlimited by any means). DSL is probably in this category, as are dialup, satelite service, and so on -- most forms of internet access. if you have dialup, your provider doesn't mind if you max out your phone line every minute that you're connected. In fact, if they had to pick, they'd rather that you use that phone line at max capacity until you're finished getting everything downloaded that you want and then disconnect, rather than downloading at half speed for twice as long. Because that way they can get by with fewer dial-in lines.
Cable modem service is different, because you do not have a dedicated pipeline from you straight to the company's core network. Instead, you share the pipeline with numerous other nodes. Bandwidth you use directly affects your neighbor's ability to get data to and from your ISP's network.
For this reason, cable modem service, while being cheaper than broadband, is also inferior. For myself, I'm sticking with my existing dialup account until I can get true broadband, something better than cable modem (such as DSL).
If I buy a laptop, there are two main questions I'll be asking:
1. How big is the display?
2. How well does it support Linux?
The first is the more important issue for me. Anything less than 16" viewable is totally useless, and I would pay a significant amount extra to get 18" viewable. I would _almost_ sell myself into indentured servanthood for 20" viewable or more.
A normal-sized keyboard would be a nice bonus too.
Yeah, I know it would take up my whole lap and weigh twice as much, but so what?
Processor? As long as it's x86-compatible, who cares? (Well, within reason. I wouldn't want a 386 laptop...)
Actually, if Apple came out with a large-screen laptop, I'd consider it. OS X isn't the _same_ as Linux, but I'd settle for it.
The dinky screens are the biggest reason I've been avoiding laptops. (There are other reasons... less standardization, more cost, poor viewing angle... but the biggest thing is the dinky screens.)
No two people get the same set of spam exactly. (I get the Asian stuff too, but I have friends who say they don't, though they do get plenty of other spam.) It all depends on where your address goes out, who harvests it, and whom they sell it to.
I have tenatively concluded that there's a ring of interrelated Asian spamhausen that share addresses with one another. They appear to have fewer addresses than the spammers elsewhere in the world but compensate by sending LOTS of mail to each address. If you get on their lists, you can expect to receive HUGE amounts of the stuff (several an hour on the same topic, sometimes). They mutate all forgeable headers on a daily basis, so the only effective ways to filter these are either by Received: header (using IP ranges -- their servers migrate across all the IP addresses they have available, and reverse lookup fails on basically all of them) or by content. I've found that it's quite effective to filter out any messages sent in character sets that I can't read anyway.
> I'm using the googlebar with Moz 1.0 (build 2002053012) > and it works just fine.
I'm using a trunk build, and am unwilling to return to the 1.0 branch because it lacks an enhancement that is very significant for me (namely, tabs close left-to-right).
But the googlebar was nifty while I had it, and I imagine the bitrot will be cleared up at some point in the future.
Slashdot Mirror
> Does anyone else see massive fragmentation
Yes. Where I work (a small public library), the word "upgrade"
means "replace with a newer system", and sometimes the old system
then gets moved to a different location and used for a different
purpose. We have the following:
1 Windows 3.11 for Workgroups (to be retired "soon")
2 Mac System 7 (yes, really)
3 Windows 95 (I think these are all OSR2)
2 MacOS 8
4 Windows 98 (I think at least three of these are SE)
3 MacOS 9
1 MacOS X (10.1.5; this is the director's workstation)
1 Windows XP / Mandrake 8.1 dual boot (this is my
workstation; I use Linux mostly but keep XP around
for testing purposes, and so I can better understand
users (patrons) who have that OS.)
1 OpenVMS 7.2-1 (this is our main catalog system)
1 Linux 2.2 (SMC TurboLinux; our low-demand cgi server.
Our main web server is hosted by our ISP, using
WebStar I believe.)
There is also a Windows PC maintained by UAN, which I
_think_ is Win98, but the library doesn't have anything
to do with upgrading/purchasing/replacing that one; it
belongs to the state's accounting network. It runs
custom software that was apparently written for DOS,
and it hangs capriciously from time to time. The only
thing I want to know is why it has an APC UPS, when the
software wedges WAY more often than the power goes out.
I would be curious to know whether fragmentation is more
extreme in small or large organisations, and whether there
is a correlation with what the organisation does (i.e.,
what the company does for its money or what the government
or nonprofit organisation does to justify its existence).
> How about adding OO?
A mistake.
Don't get me wrong; I _love_ Perl. But the way Perl 5 does OO
feels tacked on at best and really seems quite tortured. IMO,
they should have waited for the major syntax breaking in Perl 6,
which will (hopefully) allow them to do OO in a way that can
be more like a core part of the language and less like a really
strange syntax for hashes.
I certainly agree that Perl has improved since version 2, or
even since version 4, and that more improvements can potentially
be useful. If I understand correctly, early versions of Perl
5 didn't have very good language support... that surely seems
worth fixing, and I'm sure there are other things. I'd *really*
like to see the addition of textbuffers (a la elisp), which,
combined with dynamic scoping (which Perl already has), can
be extremely powerful. (Not that this is in the plans...)
Perl 6 is going to be such a huge change, it has me both
excited and scared at the same time.
> adamantium?
Funny you should mention that. I've been saying for some time
that mithril is actually just the old name for titanium steel,
and that adamantium would be something strong but more flexible,
like an alloy that includes organic matter (plasteel, anyone?),
but maybe this liquid metal stuff could be it.
> I've run Cat5, I have a couple switches and an old hub.
> I know that Cat5e and Cat6 are better than Cat 5. But
> that's about all I know
As a consumer, you don't _need_ the technical details
(unless you're just curious, in which case I'll let you
do your own web search). One thing you do need to
realise is that any given network segment is only going
to be as fast as its slowest link. A switch can make
each connected node into effectively a separate segment
(so that one slow workstation won't slow the whole
network), but a non-switching hub won't.
So, if you have a 100BaseT switching hub, that's 100Mbps,
and no cabling you can put in will make it go faster. (You
can make it go _slower_ by using less-than-cat-5 cable,
however.) If you've got a faster switch and faster cable,
you still need a faster NIC to go with it, for each system
that needs to take advantage of the speed increase.
For Perl, I want the curley braces and dollar sign in ;-)
three of the "easy" positions (where letters normally
go). Slash and backslash are negotiable.
Dvorak aside, is the difference between this layout and
QWERTY significant enough to be worth switching? (Most
people, remember, would be switching from QWERTY, not
from Dvorak.)
Personally, I doubt it, but not for the reasons you
give. In particular: with QWERTY, I can type over
50 wpm if I'm reading it off, and it's fairly standard
English (as opposed to, say, Perl). But most of the
time I don't type 50 wpm. Most of the time, I spend
time thinking of how exactly I'm going to word things.
I'm not talking about sitting idle between clauses,
but I type more slowly because I'm thinking as I go.
Typically I get somewhat less than the 50 wpm I can
physically get. (Offhand, I would guess that I tend
to get 20-35 wpm, depending on the situation, and in
special cases less. Of course, sometimes I do sit
totally idle for several seconds, but I'm excluding
those times from the average.)
For this reason, I'm not terribly interested in going
to great lengths to optimise my key layout for sheer
speed. I'm a *lot* more interested in optimising it
for comfort. In particular, reducing the amount of
strain on certain fingers that with a normal layout
frequently have to be hyperextended. The layout I'm
using at the moment has shift and ctrl on home
positions; since I adopted it, my left pinky doesn't
hurt all the time anymore. At first there was some
slowdown, and a few mistakes, but I'm beginning to
get past that. I can (under ideal, read-it-off
conditions) get 50+wpm with my new layout now. I'm
even starting to be able to switch back and forth
between my layout at home and bog-standard QWERTY at
work, with a minimal adjustment period (a couple of
minutes, typically). (It may help that I left _most_
of the keys alone, and just moved certain strategic
ones.) Having shift under a home position is to
me like having pants that fit; I'm not mostly
worried about whether I can walk faster.
> (i.e. have a 'Help' key, rather than "Press F1 for help")
.tar.gz
The VT510s where I work (a public library) are set up this
way. The catalog-only terminals (for the public) have
keys like "Search" (which is really just PF1 or PF2 or
something like that, but the software and the keytops are
designed for one another). The circulation terminals
have keys like "Patron Status" (which is really just F4
or F5 or something, but the software and the keytops are
designed for one another). Very nice. Combined with
software that quite literally tells you which buttons to
push, all but the most obstinate and thorough technophobes
have little trouble figuring out how to work the things.
The only trouble with this is, it only works so long as
your software and your keytops are made for eachother.
The minute you want to use the same computer for thirty
different applications, this will never work. Sure, you
can have certain conventions. We already have keys
reserved for page up and page down, beginning and end
of line, and such. Reserving F1 for help and labelling
it as such seems reasonable, since most apps follow that
already anyway. (The key should probably still say "F1"
in smaller print, in addition to "Help" more prominently,
in case of deviant apps.) But going much beyond that...
Hey, what about designing a keyboard with an extra row
of keys for things like Copy, Paste, and Save, and
having them generate Ctrl-C, Ctrl-V, Ctrl-S, and so
on. Sure, it wouldn't jive with Emacs[1]... But
for most end-user type applications... could be handy.
[1] NOTE: I am not dissing Emacs. (If you weren't
about to reply to what I said about Emacs, you
don't need to read the rest of this.) I would
die without Emacs, or suffer pain at any rate.
I refuse to use any computer without Emacs
installed. I use it to read my mail, as well as
usenet, and have been known to use it to browse
the web. I have written my own (fairly simple)
major mode on more than one occasion. A
of just some of my sitelisp is 161222 bytes. So
don't even start. I've compiled my own Emacs on
at least three different operating systems, in
some cases using not-yet-officially-released alpha
versions of the source, because I _wanted_ the
promised new features. I did the proof of concept
for the color-theme module. In short, I _love_
Emacs. I'm only using it as an example of something
that doesn't use the same conventions as other
applications. I'm fully aware that there are very
good reasons why it doesn't, and that people who use
Emacs don't really need other applications. And yes,
of *course* I know about remapping the keys (and have
done a lot of it, not just globally but also in mode
hooks). But I was referring to the out-of-the-box
state (not that I ever leave anything in the OOTB
state... but some people do, I guess).
> A physically-remappable keyboard could be a good thing.
> I'd probably buy one.
I have one. An Avant Stellar. Produced by Creative
Vision Technologies. It costs money, but it's a good
keyboard.
Now, the act of rearranging the keycaps doesn't cause
the layout to be remapped. But the layout can be remapped
hardwarily (or with Win9x-based software, provided), and
the (same-sized) keycaps can be swapped around. An extra
set of keytops are included for the left Ctrl and CAPS
LOCK keys, in case you want to swap them even though they
aren't the same size.
However, I wanted to swap some other stuff (put shift
under a home position...) so a few of my keys don't say
what they are. I could remedy that with white-out and
ink, if I cared...
> "Please refrain" is somewhat of a dangling clause.
... or that".
I elided the word "then" from its position as the
subordinator for the subjunctive clause. This is
standard practice in all dialects of English of
which I am aware.
> an inanimate object
My bad. (I've been fooling around with a keyboard
layout that is not strictly QWERTY, and have been
mixing up i and a lately; it'll clear up as I grow
accustomed to the layout.)
> Second: "or whether it had an unlimited capacity"
> is awkward phrasing.
Not one quarter so awkward as "whether
> Second: Ah, that's enough. I think the point is made.
My original point was that the original rant about
"moot" was stupid, and that we had better things to
discuss than grammar. Apparently I was not entirely
clear.
> The word is MOOT
Yes, the word is "moot", but if you are going to
correct grammar, and cannot manage to use correct
grammar when doing so, please refrain.
> A "mute topic" is a topic that doesn't speak.
Perhaps it would be, or perhaps it would be a topic
about which no one speaks.
> I had a partner that used to say that ALL THE
> [...] TIME [...]
No, you had a parter who used to say that frequently.
I doubt very much whether your partner was in inanimate
object, or whether it had an unlimited capacity to
continue speaking day and night.
> Sheesh, are people that ignorant and retarded???
Many people are indeed that ignorant; whether they
are retarded is a separate (and irrelevant) matter.
On a related note, it is incorrect to treble the
interogatory mark; one piece of end punctuation,
together with the introductory interjection, is quite
sufficient to indicate the mood of the sentence.
> just "Wizard" every action the user may need to take.
I *haaaaate* those things. I don't *want* to click
"Next" seventeen times.
> Considering that I never actually open many of
> the buggers, I am actually quite happy that I
> don't have to download their effing images over
> my modem connection.
But that's true regardless of your mail client's behavior.
And unless you've already _had_ the kind of privacy leak
that autoloading of remote content enables, there's no way
for the spammer to know what mail client you use, so he's
going to spam you regardless.
The real solution to _this_ problem is pre-retrieval
filtering. It is possible for a POP3 client to download
just the headers of the messages on the server and then
decide based on those which messages to retrieve in full.
Most mail clients don't do this on the grounds that it
means retrieving the headers twice for the messages that
are not excluded, but as the volume of spam increases to
more than half of all mail, it may be well worth doing.
The messages that are excluded from full retrieval by
the filter could, if desired, be left on the server for
the time being and the headers kept someplace for the
user to examine at his leisure and mark for either full
retrieval or for deletion from the server. If the user
needs to see the body to decide for sure whether it's
spam (about 0.05% of all spam plus or minus two orders of
magnitude, according to Flagrantly Estimated Statistics),
then of course he has to mark it for retrieval, but that
will be a minority case.
> If they sent all their crap inline
Plenty of them do that as well, but in that case there's
no preventable privacy issue. Spam is still bad, of course.
> Just because a particular technology is misused by idiots,
> why should we all have to do without it?
Huh? What "technology" are you talking about doing without?
The ability for your mail client to go running off to a
message-specified remote server without user action? I
suppose you also don't want to do without the "technology"
that lets your mail client execute binary attachments
without user action?
Obviously, if the message is legit, the user can always
hit the "load remote content" button, open the thing in
a web browser, or whatever.
Though I prefer just sending a URL and letting the user
doubleclick it to launch his web browser of choice, or if
his mail client doesn't support that he can copy and paste.
Spammers can do this too, but unless the user takes action
to enable it, there's no privacy leak *and* less bandwidth
consumption.
If they've replaced Linuxconf with something else, that's
almost certainly a good thing. All my experiences with
Linuxconf, across three distributions, using the text-mode
version, the Gnome version, and the web interface, have
been bad.
Not only does it leave out some of the things users often
want to configure, and obfuscate others, but it also (as
of last time I used it, which admittedly is not recently)
hangs and dumps core way too often for comfort, to say
nothing about creating inconsistent state so that the
easiest way to recover is to reinstall.
What is needed is a tool that integrates the Gnome or
KDE (whichever session the user is in) control panel
together with graphical interfaces for enabling and
disabling services (in a way that not only starts/stops
them but also changes the relevant startup scripts so
that their started/stopped-ness persists after reboot),
GUI tools for setting up multibutton mice and scroll
mice, changing resolution, colour depth, and refresh
rate of the X server, and so on. Anything you can
do from the Windoze control panel should be included.
I like the tree-on-the-left panel-on-the-right
approach of the Gnome control panel, but besides
Gnome, the rest of the system config needs to be
gathered there too -- at least, anything a newbie
might want to configure.
Dvorak makes some of the same mistakes as QWERTY,
overusing the pinkies and underusing the thumbs,
among other things. In particular:
Certain buckies (shift especially, and to a lesser extent
control) are used more frequently than the lesser-used
letters, but they are stuck out of the way where pinkies
have to be hyperextended to reach them. This is *bad*.
These keys need to be easy to reach, _especially_ since
they are often held through several consecutive keystrokes.
Only one thumb is used profitably. It would be really
nice to split that spacebar in half and use the other
half of it for something... For something you often
want to hit at the same time as other keys, such as
shift, this would be a huge improvement. In fact, if
the keyboard is split in two (as it should be, really),
there's no particular reason the left thumb can't have
a choice of two keys to hit, which could be shift and
perhaps ctrl.
Some of the seldom-used punctuation (square brackets
and curly braces, for example, which seem to pop up
mainly in Perl) are easier to reach than more often
used punctuation (such as backslash/pipe, which are
used constantly on some systems).
The Windows key on 104-key keyboards is too easy to
hit accidentally, when reaching for another key. But
hitting it disrupts whatever you are doing. It should
be up in the extreme top row, with things like PrtSc
and Pause. (Besides Windows, this also effects Gnome,
at least with some distributions.)
The extra keys between the main keyboard area and
the keypad are in the way and should be removed, or
at least moved to the outer edge beyond the keypad.
They duplicate funtionality already extant in the
keypad since 1980 if not before, in a more difficult
to use layout. (On the keypad, you can easily reach
all the cursor movement keys without moving your hand.
In the area to the left this is not the case.) Their
only purpose is to increase the distance the hand must
travel to get from the home position on the main area
(for data entry) to the home position on the keypad
(for navigation).
I'm using a modified keyboard layout, described in
another post, and it's some improvment, but I had to
work with the same _physical_ layout as everyone else,
and could have done much better given the opportunity
to physically redesign the keyboard and, for example,
make better use of the second thumb.
> What? You mean getting a key to generate a different
> character from the one that's printed on it?
The keytops on Avant keyboards are interchangeable (well,
ones that are the same size are interchangeable). That
doesn't get you a pounds symbol, but you can always use
white-out and ink to make your own keytop display. Though
if you often look at the keys while typing, you may not be
enough of a keyboardist to fully appreciate the merits of a
real quality keyboard.
The _big_ advantage of Avant keyboards is the full
remappability and programmability. I've got an Avant
Stellar, and I've got it remapped to reduce strain on my
left pinky, by repositioning the frequently-used shift and
control keys under my left and right pinkies, respectively.
The semicolon/colon (which is normally where I now have
ctrl) I moved up to the left square bracket position, and I
used the right square bracket position for backslash/pipe,
and moved the square brackets down to bottom-row positions,
since they are less frequently used. The a key (normally
where I now have shift) I moved to where k normally is
between j and l, and the less-frequently-used k I put where
left shift used to be. I repositioned the left window key
to the top function-key row, where I can find it easily but
won't hit it by mistake, turned the right-shift (which I
never use as such) into an additional backslash/pipe key,
turned the capslock (which I never use on purpose) into an
additional control key. The resulting layout takes some
getting used to, but the frequency with which I have to take
my pinky out of the home row is *vastly* reduced, which
makes for more comfort over the long term.
Of course, you may have your own set of peeves about the
standard keyboard layout. The point is, with an Avant
keyboard, you can remedy them. Oh, it does macros, too,
and with the Stellar model you get an extra set of function
keys, ripe for macroing.
One caveat: most remapping can be done with no special
software, using the hardwired key sequences, but if you want
to remap right ctrl, left shift, or 1/!, you need to use the
included remapping software, which is Win9x-based. I have a
copy of Win95 on another partition, that I was able to use
for just such an emergency; this means that I have to reboot
(before and after) each time I want to make changes to my
layout, but once I got my layout worked out I don't
anticipate frequent changes.
This full remappability/programmability is what makes CVT's
keyboards special. Yes, they're also quite good quality
physically; there are reviews out there on the web, so I
won't duplicate them here. But the versatility is the
killer feature. I was *happy* to pay more than the usual
keyboard price for this model. CVT is for keyboards what
Lian Li is for cases. If you don't think keyboards are a
big deal, if you're satisfied with your $40 ergonomic
keyboard (or even with a $9 cheapie), then an Avant may not
be for you. If you find yourself dissatisfied with the
shortcomings of traditional keyboard layouts, disgusted with
the quality of membrane keyboards that succumb to humidity
and have to be replaced frequently, in general, if you just
want a better keyboard, I see an Avant model in your future.
(I'm just a satisfied customer; I have no affiliation with
the company.)
The test: if you are looking at a webpage, and need to scroll, do you
A) Reach for the mouse.
B) Use the mouse without needing to reach, because your
hand was on it already.
C) Keep your hands on the keyboard and use the relevant
keys to scroll, because they're closer and easier to
reach than the mouse.
If you answered C, you will appreciate a quality keyboard.
If you answered B, you won't care. If you answered A, you
are a borderline case, and may or may not find a quality
keyboard worthwhile, depending on your budget.
> anyone who says mandrake is only for beginners
Not "only for beginners" per se, but Mandrake is especially
well-suited for users crossing over from other OSes, especially
from Windows. This doesn't mean it can't be used for other
things, only that the places where it shines out and is
notably better than other distros are mostly to do with
being friendly to the unwashed masses. Yes, it's also
optimized for Pentium... but in that regard it can't
claim to be better than Gentoo, for example; Mandrake
is, however, more friendly to newbies from Windozeland
than Gentoo. That's the most significant strength of
Mandrake (and it is an important thing).
Second-tier is not a _bad_ thing. But yes, I'd say
Suse is a second-tier distro. (So is Mandrake, as
far as that goes.)
Overall, there are probably only two first-tier Linux
distros (Debian and RH), as I estimate it. What makes
them first-tier is that they form the foundation for
other distributions that are based on them. So the
second-tier and third-tier distros fall into three
categories: those that are based on Debian, those
that are based on RH, and those that go their own way.
(These can be further subdivided, of course.)
I'm not a Suse user, but my understanding is that they
go their own way, not tracking closely with either
RH or Debian. Which is fine. But unlike RH or
Debian, I don't know of any distros that are based
on Suse.
United Linux seems to me to be several second-tier
distros getting together to form a third first-tier
distribution. Now there will be numerous distros
that can be classified as UL-based, as well as those
that are RH-based and those that are Debian-based,
and of course those that go their own way (e.g.
Gentoo -- and speaking of Gentoo, I ran across the
word "gentoo" in a book yesterday (_Endurance:
Shackleton's Incredible Voyage_), wherein it seems
to be a type of penguin.)
I don't have a problem with that. As the popularity
of Unix in general and Linux in particular continues
to grow (as has been happening nonstop since the
original development of Unix at Bell Labs), there
will be room for more first-tier distributions,
as well as more second-tier distributions, more
third-tier distros (yes, the line between second-tier
and third-tier is probably pretty blurry), and more
specialty distributions as well, to say nothing of
more bundled-with-hardware distributions.
For now there are few enough systems sold with
Linux-based distros preinstalled that they use
existing distros; when it really catches on, the
hardware vendors will put their own distros together,
just like the bizzarroid bundling they've been doing
all along with other kernels (both unices and other).
Like with all unices, they'll have more freedom
(than with, say, MS OSes) to make the branding
really pervasive, with their corporate logo not just
on the default wallpaper, but built into the web
browser (which has its start page set to their
corporate site), embedded in the panel background,
all over the GUI administrative apps (think:
"Compaq Control Center"), all over the bundled
office suite, the bundled email client, the
bundled music/video player, and so on and so
forth. Think of the Gnome foot or the KDE
gears being replaced with the hardware vendor's
logo. Think of the default signature in the
bundled email client advertising the vendor,
unless and until the user changes it.
There are huge opportunities for hardware vendors
here. When the software is a commodity, you can
brand it any way you want, and you don't have to
share your branding with another company. Sun
and Apple have been doing this for a while, with
more expensive hardware, but there's no reason
it can't be done with low-end hardware and by
other vendors.
> Most spam I get these days is HTML phone-home style. As soon
> as I read it, it's off requesting images from somewhere.
If your mail client lets it do this, your mail client is
br0ken. Good mail clients NEVER retrieve content from
elsewhere on the internet during the course of merely
reading a message. That's a fundamental privacy flaw,
a serious performance issue, and in certain cases even
a security hole (though not with mere images). There is
also no legitimite reason for it in a normal message.
(If you *must* have inline images (a highly dubious
point; attachments will do for way more than 99% of
cases), that's what the disposition/inline MIME thingy
is for.)
> > For myself, I'm sticking with my existing dialup
> > account until I can get true broadband, something
> > better than cable modem (such as DSL).
>
> What a ridiculous conclusion.
I did say "for myself". Also note that I'm considering
the costs of switching service providers (e.g., tracking
down hundreds of people and getting them all my new
email address, tracking down all the search engines
with my webspace indexed and getting them to index
the space at the new provider, and so on), and when
all of that is thrown in, cable modem service (at
least in this area) is just not good enough to warrant
switching, for me.
Also, several repliers have cited impressive download
speeds for cable modem service, but if I went broadband
I'd want bandwidth in the other direction, too...
What I _really_ want is a residential T1...
Not all DSL is created equal, but you already knew that.
> When you are talking viewable, do you mean in reference
> to the laptop (ie LCD) world, or the desktop (CRT still,
> for the most part) world? I am confident you know the
> difference.
I understand the difference. I said viewable, because
my current monitor is 18 inch viewable (a "19 inch" CRT),
and if I got a laptop for my next system, I'd prefer not
to lose screen real estate. That's why I chose 18,
instead of 17 or 19, as the value for a future laptop
to match in order to be an acceptable replacement for
my current (desktop) system.
> If Linux is such a priority,
Not so much Linux specifically as Unix in general, but
I said Linux because it tends to have better hardware
driver support than (say) BSD, particularly on low-end
consumer-type hardware.
> and Linux is still quite a bit of command line, or
> maybe you write a decent amount of code, than I am
> not sure I understand you wants/desires. Gaming?
I'm not sure I understand the question. I was talking
about being able to get all the hardware (NIC, sound
if possible, and so on) working with the a Linux kernel.
Hardware that doesn't work except with Windoze or NT
is of substantially decreased value to me.
> And your gripe about poor viewing angles has been
> a moot point for at least a year or two now.
Such has not been my experience. Current LCD screens
(at least, the low-end ones) do not have viewing
angles that impress me much. In particular, if you
move your head as little as thirty degrees (relative
to the screen) it changes the colour ballance. It's
not as horribly bad as it once was, but LCDs still
need some improvement to catch up to CRTs.
Still, I'm not a graphics artist, and so to me,
screen _size_ is more important than colour accuracy,
so I might be able to settle for a laptop screen if
it's large enough.
15.7" is really not large enough. I could use maybe
use it on trips, but I can't afford to buy both a
laptop (for trips) _and_ a new desktop for regular
use. Plus, it would drive me out of my mind not
having my regular (heavily customised) environment
on the laptop. So if I'm going to get a laptop,
I'd really want one with a display large enough
that I could also use it as my regular system. So,
18" or thereabouts.
> I wonder if DSL has different rules?
I suspect it does. Actually, DSL has the normal rules;
it's cable modem service that has different rules.
There are good reasons for this, and it comes down
to design. The cable network was _designed_ to
deliver lots of the SAME content to every subscriber.
Yeah, it can carry 50+ channels, but it takes the
same 50+ channels to everybody. Need to add another
thousand users? No problem: just tee off the line
that's already there. As cable TV grew, they've
been teeing the lines for years.
As a result, the lines can handle pretty substantial
bandwidth to an individual node, but the _total_
amount of bandwidth on a given subnet has some pretty
stiff limits on it. In particular, the total amount
of available bandwidth, that has to be split between
all nodes, is not necessarily greater than what one
node is capable of receiving (unless they place some
artificial cap on the latter).
This is backwards from almost every other network.
If you have, say, a T1 line, your provider (typically)
won't whine much if you use its maximum capacity,
because the T1 line itself is the chokepoint. At
the other end is a router that's probably on your
ISP's core LAN (for small ISPs) or core WAN segment
for your area. The bandwidth there is somewhat more
plentiful (though not unlimited by any means). DSL
is probably in this category, as are dialup, satelite
service, and so on -- most forms of internet access.
if you have dialup, your provider doesn't mind if you
max out your phone line every minute that you're
connected. In fact, if they had to pick, they'd
rather that you use that phone line at max capacity
until you're finished getting everything downloaded
that you want and then disconnect, rather than
downloading at half speed for twice as long. Because
that way they can get by with fewer dial-in lines.
Cable modem service is different, because you do
not have a dedicated pipeline from you straight to
the company's core network. Instead, you share
the pipeline with numerous other nodes. Bandwidth
you use directly affects your neighbor's ability
to get data to and from your ISP's network.
For this reason, cable modem service, while being
cheaper than broadband, is also inferior. For
myself, I'm sticking with my existing dialup
account until I can get true broadband, something
better than cable modem (such as DSL).
YES.
If I buy a laptop, there are two main questions I'll
be asking:
1. How big is the display?
2. How well does it support Linux?
The first is the more important issue for me. Anything
less than 16" viewable is totally useless, and I would
pay a significant amount extra to get 18" viewable. I
would _almost_ sell myself into indentured servanthood
for 20" viewable or more.
A normal-sized keyboard would be a nice bonus too.
Yeah, I know it would take up my whole lap and weigh
twice as much, but so what?
Processor? As long as it's x86-compatible, who cares?
(Well, within reason. I wouldn't want a 386 laptop...)
Actually, if Apple came out with a large-screen
laptop, I'd consider it. OS X isn't the _same_
as Linux, but I'd settle for it.
The dinky screens are the biggest reason I've been
avoiding laptops. (There are other reasons... less
standardization, more cost, poor viewing angle...
but the biggest thing is the dinky screens.)
No two people get the same set of spam exactly. (I get the Asian
stuff too, but I have friends who say they don't, though they do
get plenty of other spam.) It all depends on where your address
goes out, who harvests it, and whom they sell it to.
I have tenatively concluded that there's a ring of
interrelated Asian spamhausen that share addresses with
one another. They appear to have fewer addresses than
the spammers elsewhere in the world but compensate by
sending LOTS of mail to each address. If you get on
their lists, you can expect to receive HUGE amounts of
the stuff (several an hour on the same topic, sometimes).
They mutate all forgeable headers on a daily basis, so
the only effective ways to filter these are either by
Received: header (using IP ranges -- their servers
migrate across all the IP addresses they have available,
and reverse lookup fails on basically all of them) or
by content. I've found that it's quite effective to
filter out any messages sent in character sets that
I can't read anyway.
> I'm using the googlebar with Moz 1.0 (build 2002053012)
> and it works just fine.
I'm using a trunk build, and am unwilling to return to the
1.0 branch because it lacks an enhancement that is very
significant for me (namely, tabs close left-to-right).
But the googlebar was nifty while I had it, and I imagine
the bitrot will be cleared up at some point in the future.