I'm wondering what happens to the hard links when rsync decides it only needs to update part of a file.
I wondered about the same thing too. Given he reports permissions getting changed for older versions, I wonder if the contents of older copies also get changed.
I do my space efficient multiple levels of (online) backups the other way around: rsync the backup into a snapshot (not a true snapshot, but close enough on the systems I do it on), and then run a script that compares the previous snapshot and the new snapshot and links identical (contents, permissions, etc) files together.
I also generally keep a current directory which has an exact rsync copy of the thing being backed up, and back that up to tape regularly -- the snapshots are just there for easy online recovery. Doing this has greatly reduced the need to recover from tape, but it is nice to know the option to do so is still available.
I've ranted about this before, but I have to say it again: Programming is taught ass-backwards in college. Assembly language should be the FIRST thing taught, and then gradually building up to higher and higher levels of abstraction.
I go back and forth on my thinking as to which order things should be taught in: sometimes I feel that assembly language is an absolute must as a starting point (preferable, after some hardware design), and sometimes I think a birds eye view of a computer as an abstract high level machine is an essential place to start. (It generally depends on which set of mistakes I've seen someone making most recently, I think!)
But irrespective of that, IMHO the most productive, talented, progammers (and system administrators, and....) have a good insight into the machine at multiple levels, ideally spanning all the way from some idea about the hardware through CPU internals, through assembly language, through portable assembly language (languages like C that are close to the hardware), through high level languages. People completely missing one or more of those levels seem to overlook something, particularly when debugging.
And often if they don't have those multiple layers of experience, they act like their reality is broken when something goes wrong. Abstractions (like high level languages) are nice when they work, but sometimes the lower levels still shine through and catch you out. If you're aware (even vaguely) of the lower levels being there and roughly how they work, then the lower levels shining through is just another debugging hint, instead of a shock to your reality.
I'm still curious as to how he's going to change Ad-Aware to prevent it being uninstalled by this other program. Does anybody know?
This calls to mind the old story of Robin Hood and Friar Tuck. Essentially instead of having one program that can be killed off/removed, you have two programs each keeping an eye on the other, and starting/reinstalling the other as required.
As someone commented in the last thread on this topic, this all rather reminds me of Core Wars, played out at large. We just need a better way of keeping score...
You've been paying all along, and today, you were simply given an option to pay in a different way, replacing the old way. And if you don't want to change, you don't have to. Why is this a problem?
The change is from not-particularly-annoying relatively-small banner ads to, apparently (from the way it was introduced) annoying/distracting/in-your-face block ads. Maybe we have been paying all along; but the price just got a whole lot more expensive.
The way the more-advertising/pay-to-avoid-it system was announced suggests that they're essentially going to make it so that people will want to pay for a subscription to avoid the content-keeping-the-ads-apart effect (or to abandon the site for somewhere else).
Both (people leaving, and people paying a subscription and worrying about using it up) will affect Slashdot. For the worse. The very thing that brought the audience (and hence the advertising revenues) will be discouraged/driven away.
I was merely trying to suggest a way that the things that make Slashdot good could be preserved while still allowing Slashdot a way to get money to pay for all the costs of running the site. From reading other comments here, it seems I'm not alone in suggesting a reward-for-contributing type system.
This change will change the whole site dynamics. For the worse, I think, in its current form.
Slashdot offers two main things:
A clipping service (the front page, etc)
A reasonably workable discussion forum for comments on the each article, that allows "good" comments to be seen fairly easily
Both of these things rely heavily on "community involvement". Most of the links for the clipping service come from contributions; all the discussion, and all the filtering of the discussion (moderation) comes from the community.
People got rewarded for sending in link suggestions with their name in lights; people got rewarded for good posts with karma; people got rewarded for moderation/meta-moderation with (some) karma. The efforts/rewards were reasonably well balanced to produce the current Slashdot.
Now there's a new factor. Annoying adverts. (I'm assuming they'll be annoying because of the way this is approached, the "we know you won't like this, so here's a way you can buy your way out of it" approach.)
Which changes the whole dynamics of the site. Suddenly people get "charged" for seeing their name in lights (with annoying adverts, or actual money). Suddenly people get "charged" for reading the comments so they can post. Suddenly people get "charged" for reading the comments so they can moderate them. And perhaps people even get "charged" for reading moderations so they can do meta-moderation. Incentives not to do these things. These things which make Slashdot what it is now.
If Slashdot wants to make a major change like this, and not dramatically change the "feel" of Slashdot, then it needs to be made balancing these contributions/rewards. Sending in article links needs to be rewarded; posting good comments needs to be rewarded; doing moderation and meta-moderation needs to be rewarded. In the context of the new change.
Some things Slashdot should consider:
Having an article link posted to the front page/a section should be rewarded by some number of "advertising free" pages. 250/500/1000 page views, perhaps based on interest generated in it.
(Click through counting may be required; I'm surprised click-through counting isn't done already.)
Posting a really good comment, say one that is moderated to 5 AND all the moderations are supported by "that's right" meta-moderations should be rewarded. 100/200/300 page views, say.
Moderation done well (supported by meta-moderators) should be rewarded. 25/50/75 page views, say, for the whole set of (5) moderations.
Meta-moderation done well (same opinion as other meta-moderators) should be rewarded; say 5 page views for the whole set of (10) meta-moderations if they're all supported.
Without these sorts of balancing rewards all the things that make Slashdot good will be discouraged by annoying adverts (persuading people to go elsewhere), or by the knowledge that if you load the comments to contribute/moderate it's going to cost you, so why bother.
I've no problem with contributing to Slashdot, even money if the framework for the contribution is right (the current scheme is not). But all the contributions which make Slashdot what it is need to be recognised in the new framework.
Distributing power over fibre optics, and already in the eighties ?? That's very advanced.
Some of the article suffers from "string the buzzwords together" issues. I think that's one of them.
I think in that instance it was more an observation that network connections could be laid out in a grid (rather than point to point), like the power network is laid out. So losing one connection didn't lose everything. Richard Naylors's background (further back than his job at the City Council) is in electricity distribution.
Citylink (largely pushed by Richard Naylor) have done some neat things with powering devices over Cat-5, so that they can have routers with just a single lead going to them. (Ideal for phone-pole based switches and routers for instance.) Which almost takes us the whole circle back to powering things over a network connection...
Since they consider themselves a LAN, I was wondering what addressing scheme they're using.
Are they using "real" IP addresses? If so, what class?
Citylink offers two main services. One is a "dark fiber" connection (they put it in, manage it for you) which lets you do whatever you like with it. This is the equivilent of a leased line from a telco between two buildings you own (or you and a client). You can use any addresses, protocols, etc, on that without affecting anyone else.
The other service is their public MAN. All the ISPs in New Zealand (all the significant ones with any presence in Wellington anyway) are connected to this Citylink public MAN. To use it for Internet access you go to one of these ISPs and get some addresses assigned for you to use. Because it's a layer-2 network, all these addresses from different ISPs can be used in parallel without affecting each other. (Just like you can on a LAN segment for testing, etc.)
The really big win of the public MAN and all the ISPs being connected is that changing ISP is pretty easy if you need to -- you just need new IP addresses (for CIDR allocations), or a new set of routing entries (for those with real address space of their own). Makes it a lot cheaper, and easier, than having to get new leased lines run, etc.
Yes, it's strictly a MAN. Richard Naylor, the man with the vision behind this, had been talking about the idea of a MAN for Wellington for many years. I remember at one meeting when he talked about the idea of a MAN, and commented on the "gender issues" in the name, someone (Paul Gillingwater, IIRC; hi Paul), suggested it could be called WOMAN -- Wellington's Own Metropolitian Area Network
While Richard was pushing this during his time at the City Council, Citylink (the network) didn't really get created until Citylink the company was created. It's a privately owned company (owned by various ISPs, Telcos, and other people) that started with about NZ$50,000 (approximately US$25,000). It's amazing what they've been able to do starting with so little.
It's an amazing service to have available in the city, and priced very competitively with trying to get traditional leased lines from a telco. With the added advantage of being able to change ISP easily (just get some new IP addresses assigned; all the ISPs are on Citylink), it's a big win.
Thank you Richard for pushing for the vision to come true.
So I'm no closer to knowing if I win4lin, for example, would be overall faster (as they claim) than vmware which I currently own (well, license, but I paid, damnit). I very well may shell out another $79 if something like win4lin is significantly faster.
Between my company and myself I own two copies of VMWare (one VMWare workstation, one VMWare Express -- VMWare Express is basically VMWare Workstation, but configured to run only Windows 95/98, and much cheaper), and two copies of Win4Lin.
Windows 98 in VMWare on a Dual Pentium Pro/200 with the special drivers, run full screen, was usable for interactive things (Word, Excel, etc), but too slow without the special drivers. It was clearly much "slower than the real thing" too. Windows 98 in VMWare on an Athlon 900 is very usable; although certainly "slower than the real thing". (The Dual PPro is now being used as a test server, running various VMWare instances; the flexibility of doing that makes up for the speed loss.)
Windows 98 in Win4Lin is very usable on my Pentium 200 laptop, more so than VMWare was on a Dual Pentium Pro/200. Windows 95 (OSR2)
in Win4Lin is also very usable remotely displayed to a diskless workstation (Pentium 200-based, over a dedicated 100Mbps Ethernet link), from a Duron 700 system.
I'd guesstimate that Win4Lin gives you around 90% of the performance of Windows native on the hardware for many things, and VMWare gives you something like 50% of the native performance. I've run Windows natively on my laptop (its configured to dual boot), and didn't find Windows in Win4Lin noticably too slow for anything I did (well not much more so than natively: big compile jobs took ages!). But I'd never try running VMWare on that Pentium 200 laptop.
If you can live with Win4Lin's restrictions
(less emulation of hardware, more merging of things, business applications focus, need for a patched kernel), then I'd definitely recommend it where speed is an issue. But for the things I do I have uses for both Win4Lin and VMWare, for slightly different tasks.
FWIW, I've been using Win4Lin 1.0 and Win4Lin 2.02; I've heard that Win4Lin 3.0 is at least as good (from someone who has used all three), but haven't yet upgraded myself. Only Win4Lin 3.0 is supported with the recent kernel patches though, so I'll need to upgrade soon. Finally both Win4Lin and VMWare benefit from having lots of memory available; VMWare needs more memory to get to its sweet spot than Win4Lin does though.
The most obvious change is that now every single page sends a brand new cookie to accept/reject (unlike the old code which sent one only when you logged in or didn't supply a cookie). This is one of my pet hates on websites -- being bugged by new cookies with every single page -- and rapidly makes an otherwise good website too annoying to bother with.
Surely with a one year expiry time on the cookies it is only necessary to send them once a month or so at the most? Or perhaps this is the Slashcode version of Chinese Water Torture.
Slashdot Mirror
I wondered about the same thing too. Given he reports permissions getting changed for older versions, I wonder if the contents of older copies also get changed.
I do my space efficient multiple levels of (online) backups the other way around: rsync the backup into a snapshot (not a true snapshot, but close enough on the systems I do it on), and then run a script that compares the previous snapshot and the new snapshot and links identical (contents, permissions, etc) files together.
I also generally keep a current directory which has an exact rsync copy of the thing being backed up, and back that up to tape regularly -- the snapshots are just there for easy online recovery. Doing this has greatly reduced the need to recover from tape, but it is nice to know the option to do so is still available.
Ewen
I go back and forth on my thinking as to which order things should be taught in: sometimes I feel that assembly language is an absolute must as a starting point (preferable, after some hardware design), and sometimes I think a birds eye view of a computer as an abstract high level machine is an essential place to start. (It generally depends on which set of mistakes I've seen someone making most recently, I think!)
But irrespective of that, IMHO the most productive, talented, progammers (and system administrators, and....) have a good insight into the machine at multiple levels, ideally spanning all the way from some idea about the hardware through CPU internals, through assembly language, through portable assembly language (languages like C that are close to the hardware), through high level languages. People completely missing one or more of those levels seem to overlook something, particularly when debugging.
And often if they don't have those multiple layers of experience, they act like their reality is broken when something goes wrong. Abstractions (like high level languages) are nice when they work, but sometimes the lower levels still shine through and catch you out. If you're aware (even vaguely) of the lower levels being there and roughly how they work, then the lower levels shining through is just another debugging hint, instead of a shock to your reality.
Ewen
This calls to mind the old story of Robin Hood and Friar Tuck. Essentially instead of having one program that can be killed off/removed, you have two programs each keeping an eye on the other, and starting/reinstalling the other as required.
As someone commented in the last thread on this topic, this all rather reminds me of Core Wars, played out at large. We just need a better way of keeping score...
Ewen
The change is from not-particularly-annoying relatively-small banner ads to, apparently (from the way it was introduced) annoying/distracting/in-your-face block ads. Maybe we have been paying all along; but the price just got a whole lot more expensive.
The way the more-advertising/pay-to-avoid-it system was announced suggests that they're essentially going to make it so that people will want to pay for a subscription to avoid the content-keeping-the-ads-apart effect (or to abandon the site for somewhere else).
Both (people leaving, and people paying a subscription and worrying about using it up) will affect Slashdot. For the worse. The very thing that brought the audience (and hence the advertising revenues) will be discouraged/driven away.
I was merely trying to suggest a way that the things that make Slashdot good could be preserved while still allowing Slashdot a way to get money to pay for all the costs of running the site. From reading other comments here, it seems I'm not alone in suggesting a reward-for-contributing type system.
Ewen
This change will change the whole site dynamics. For the worse, I think, in its current form.
Slashdot offers two main things:
Both of these things rely heavily on "community involvement". Most of the links for the clipping service come from contributions; all the discussion, and all the filtering of the discussion (moderation) comes from the community.
People got rewarded for sending in link suggestions with their name in lights; people got rewarded for good posts with karma; people got rewarded for moderation/meta-moderation with (some) karma. The efforts/rewards were reasonably well balanced to produce the current Slashdot.
Now there's a new factor. Annoying adverts. (I'm assuming they'll be annoying because of the way this is approached, the "we know you won't like this, so here's a way you can buy your way out of it" approach.)
Which changes the whole dynamics of the site. Suddenly people get "charged" for seeing their name in lights (with annoying adverts, or actual money). Suddenly people get "charged" for reading the comments so they can post. Suddenly people get "charged" for reading the comments so they can moderate them. And perhaps people even get "charged" for reading moderations so they can do meta-moderation. Incentives not to do these things. These things which make Slashdot what it is now.
If Slashdot wants to make a major change like this, and not dramatically change the "feel" of Slashdot, then it needs to be made balancing these contributions/rewards. Sending in article links needs to be rewarded; posting good comments needs to be rewarded; doing moderation and meta-moderation needs to be rewarded. In the context of the new change.
Some things Slashdot should consider:
Without these sorts of balancing rewards all the things that make Slashdot good will be discouraged by annoying adverts (persuading people to go elsewhere), or by the knowledge that if you load the comments to contribute/moderate it's going to cost you, so why bother.
I've no problem with contributing to Slashdot, even money if the framework for the contribution is right (the current scheme is not). But all the contributions which make Slashdot what it is need to be recognised in the new framework.
Ewen
Some of the article suffers from "string the buzzwords together" issues. I think that's one of them.
I think in that instance it was more an observation that network connections could be laid out in a grid (rather than point to point), like the power network is laid out. So losing one connection didn't lose everything. Richard Naylors's background (further back than his job at the City Council) is in electricity distribution.
Citylink (largely pushed by Richard Naylor) have done some neat things with powering devices over Cat-5, so that they can have routers with just a single lead going to them. (Ideal for phone-pole based switches and routers for instance.) Which almost takes us the whole circle back to powering things over a network connection...
Citylink offers two main services. One is a "dark fiber" connection (they put it in, manage it for you) which lets you do whatever you like with it. This is the equivilent of a leased line from a telco between two buildings you own (or you and a client). You can use any addresses, protocols, etc, on that without affecting anyone else.
The other service is their public MAN. All the ISPs in New Zealand (all the significant ones with any presence in Wellington anyway) are connected to this Citylink public MAN. To use it for Internet access you go to one of these ISPs and get some addresses assigned for you to use. Because it's a layer-2 network, all these addresses from different ISPs can be used in parallel without affecting each other. (Just like you can on a LAN segment for testing, etc.)
The really big win of the public MAN and all the ISPs being connected is that changing ISP is pretty easy if you need to -- you just need new IP addresses (for CIDR allocations), or a new set of routing entries (for those with real address space of their own). Makes it a lot cheaper, and easier, than having to get new leased lines run, etc.
Yes, it's strictly a MAN. Richard Naylor, the man with the vision behind this, had been talking about the idea of a MAN for Wellington for many years. I remember at one meeting when he talked about the idea of a MAN, and commented on the "gender issues" in the name, someone (Paul Gillingwater, IIRC; hi Paul), suggested it could be called WOMAN -- Wellington's Own Metropolitian Area Network
While Richard was pushing this during his time at the City Council, Citylink (the network) didn't really get created until Citylink the company was created. It's a privately owned company (owned by various ISPs, Telcos, and other people) that started with about NZ$50,000 (approximately US$25,000). It's amazing what they've been able to do starting with so little.
It's an amazing service to have available in the city, and priced very competitively with trying to get traditional leased lines from a telco. With the added advantage of being able to change ISP easily (just get some new IP addresses assigned; all the ISPs are on Citylink), it's a big win.
Thank you Richard for pushing for the vision to come true.
Between my company and myself I own two copies of VMWare (one VMWare workstation, one VMWare Express -- VMWare Express is basically VMWare Workstation, but configured to run only Windows 95/98, and much cheaper), and two copies of Win4Lin.
Windows 98 in VMWare on a Dual Pentium Pro/200 with the special drivers, run full screen, was usable for interactive things (Word, Excel, etc), but too slow without the special drivers. It was clearly much "slower than the real thing" too. Windows 98 in VMWare on an Athlon 900 is very usable; although certainly "slower than the real thing". (The Dual PPro is now being used as a test server, running various VMWare instances; the flexibility of doing that makes up for the speed loss.)
Windows 98 in Win4Lin is very usable on my Pentium 200 laptop, more so than VMWare was on a Dual Pentium Pro/200. Windows 95 (OSR2) in Win4Lin is also very usable remotely displayed to a diskless workstation (Pentium 200-based, over a dedicated 100Mbps Ethernet link), from a Duron 700 system.
I'd guesstimate that Win4Lin gives you around 90% of the performance of Windows native on the hardware for many things, and VMWare gives you something like 50% of the native performance. I've run Windows natively on my laptop (its configured to dual boot), and didn't find Windows in Win4Lin noticably too slow for anything I did (well not much more so than natively: big compile jobs took ages!). But I'd never try running VMWare on that Pentium 200 laptop.
If you can live with Win4Lin's restrictions (less emulation of hardware, more merging of things, business applications focus, need for a patched kernel), then I'd definitely recommend it where speed is an issue. But for the things I do I have uses for both Win4Lin and VMWare, for slightly different tasks.
FWIW, I've been using Win4Lin 1.0 and Win4Lin 2.02; I've heard that Win4Lin 3.0 is at least as good (from someone who has used all three), but haven't yet upgraded myself. Only Win4Lin 3.0 is supported with the recent kernel patches though, so I'll need to upgrade soon. Finally both Win4Lin and VMWare benefit from having lots of memory available; VMWare needs more memory to get to its sweet spot than Win4Lin does though.
The most obvious change is that now every single page sends a brand new cookie to accept/reject (unlike the old code which sent one only when you logged in or didn't supply a cookie). This is one of my pet hates on websites -- being bugged by new cookies with every single page -- and rapidly makes an otherwise good website too annoying to bother with.
Surely with a one year expiry time on the cookies it is only necessary to send them once a month or so at the most? Or perhaps this is the Slashcode version of Chinese Water Torture.