uh uh... you'll have to tell me, and I think you already did: You like the code and you like the people. I think that's evidence enough that you're home at last.
Besides, when they haven't thrown out the rowdy likes of me, there's not much chance they won't welcome you:)
This is a bit of a troll for a Holy War, don't you think? Obviously, and I say that emphatically: obviously every single one of the open source and free software CMS exist for a reason, because unlike proprietary software, people don't sit down and simply imagine that mice will beat a path to their door if they invent the perfect content system... these things get written because someone needed them, each feature gets added because someone needed them, and for those someones, the software is a priori appropriate and useful. QED.
Now, on to a more interesting question: Do you really want "GNU/opensource" systems? I question that because of the way you ask for a system that will be complete for your purposes instead of asking the more sane and reasonable question, "in which of these communities would I best fit in"
Media Girl, with a handle like that you should know that media is something between people -- Free software isn't about supplying you with the perfect glass of gratuitous beer, it's about people working together, co-operatively, optimally working to distribute the considerable load of re-inventing shared solutions so that each can better concentrate on their specific needs -- if you can't get along with people in this way, then my advice is go buy some nice proprietary kit with a handy 1-800 number you can yell at when your expectations aren't met.
On the other hand, let's say you simply had a bad choice of words and you really do plan to participate in the crafting of the software you need... then you can safely pick your CMS with a blindfold and a pin because the real question you want to ask is social, and much as some think otherwise, we can't answer social questions on SlashDot, you just have to go to the party, talk to people and find out for yourself where you'd be happier among the GeekLoggers, B2's or among the MT-Pluggers.
For my own use, I wandered from party to party for a long time, spending at least a few months in each while I met people, saw where things were going, got a sense of whether or not these were the sorts of people I could get along with and the sort of project where I wanted to be involved.
I started with my own PMTS -- bad move, too much trouble doing everything yourself, way to costly for way too little result. So I set out to find a shared-solution...
MetaDot -- almost a proprietary lock on the development, lets in no input, offers no control over their development, and then they took a direction that didn't match my own leaving me stranded
GeekLog -- very clever kids, but kids mostly, and never could figure out who was who or who did what or where it might be archived.
B2 -- had to hack it into B2++ and gave up because no one else was interested in the ++ so I was back to the same self-supplier; ditto with something that I think was called Xoop, nice code, but I'd be on my own, which is where I started
MT -- elegant although very buggy and doesn't scale very well (I write a lot), I hung on until Mena called us all 'thieves' for actually using her 'free' software; then I got real pissed off and dumped all my use of their stuff. Having finally learned my lesson, I went out in search of real 'free software' where I could actually participate.
There was a long string of others -- I chronicalled several of them in an article/thread on advogato.org. I eventually settled at Drupal.org where the code maybe had warts (fewer all the time) and the code isn't as OO or as standards-based as maybe I'd have preferred (in other parts, like Conditional-GET, it's way too standards-based), but their archives are clean and orderly, the participatory environment is excellent (they use Drupal to build Drupal) and the core troupe are welcoming, eager, intelligent and open to new participation.
Your needs may be different, but that's enough for me.
Please consider this an innocent question, not an attack, because I truly want to know: If this method is so great and your method is so much better, why are none of these methods common in modern email or MTA software?
Actually, there's two, and both are easily found by simply entering "bayesian spam" into Google:
There already exists a generic method (ifile) to fold this technique into a procmail script, which means you don't need any special-purpose email program; any company that thinks its going to replace the email-browser is dreaming. I also downloaded some experimental code for Emacs GNUS, but it was too clunky for anything more than a demonstration of the rating method. Since ifile works with procmail, any ISP could use it to tag suspect email, so it wouldn't matter if there are both geeks and sex-starved teens (not the same thing?) in the audience; each can do with the extra tags as they wish.
the Google results also show this is not a new problem; research, heavy research, has been applied to Bayesian network classification of spam emails since 1994... so it is the first approach, or close to it. My first question is then, "if it works so well, with 0% false positives, then why did everyone, even Microsoft, abandon it?" IFile has been around for a long time, yet none of even the Linux distros include it by default. That's a little suspicious, don't you think? If the method is so foolproof, why are there no fools using it?
Excuse my greying cynicism, but there's no mention in Paul's paper of how he's accounting for the mass failure of the corpus of work that goes before him, and I get a little dubious when one lone programmer claims they can out-think large numbers of trained professionals and academics. Yes, it does happen, but when you hear hoofbeats in the street, it's usually not a zebra.
Please feel free to enlighten me about the above two; I'm not investing in Paul's employer, so the first issue is not nearly as important as the second, but as a spam-victim, I truly do so want to believe there's a magic anti-spam bullet, I just have trouble believing this particular story based on the data at hand.
For those having trouble with the Javascript/mime-detect stuff, here's a hack tip: You will notice that the URL is an HTTP GET request where the parameter is url=http... -- that's the interface to the javascript process; if you'd like to by-pass this (and if you have Java WebStart already installed) just replace the url= with jnlp= and the Gateway will send you the JNLP file directly.
For those who want something useful
on
Swarmcast GPLed
·
· Score: 1
Don't like the sample content? Why not register your own, for example, how about RedHat 7.1?...
Well, for one thing, the per-block SHA hash of the trojan wouldn't match the original, so the swarm would learn to ignore that source of data. At most, you slowed down a few people imperceptably while they fetched useless blocks from your trojan server.
Re:Hello, thats the wrong license
on
Swarmcast GPLed
·
· Score: 1
Not quite, but still the wrong license: He's quoting the license for the Gateway, which is the website that acts as a peer-introduction-service; the Gateway is not GPL (it's a very specific thing, not a very general tool or protocol).
Don't worry, the Gateway isn't tracking you: Your node anonymously requests a "lease" on a resource and until that lease expires, new mesh members are given your node as a possible entry-point. It's a salve until multicast discovery can do the same thing in pure peer-to-peer mode. Other apps such as Ximian, where the audience is focussed on some website, will probably also use Gateway-like software to stitch the mesh together.
But don't confuse them: There is precious little SCL code in the Gateway; it is only there to keep a temp cache of REMOTE_HOST headers and link them to known resource SHA hashed IDs.
Slashdot Mirror
uh uh ... you'll have to tell me, and I think you already did: You like the code and you like the people. I think that's evidence enough that you're home at last.
:)
Besides, when they haven't thrown out the rowdy likes of me, there's not much chance they won't welcome you
Now, on to a more interesting question: Do you really want "GNU/opensource" systems? I question that because of the way you ask for a system that will be complete for your purposes instead of asking the more sane and reasonable question, "in which of these communities would I best fit in"
Media Girl, with a handle like that you should know that media is something between people -- Free software isn't about supplying you with the perfect glass of gratuitous beer, it's about people working together, co-operatively, optimally working to distribute the considerable load of re-inventing shared solutions so that each can better concentrate on their specific needs -- if you can't get along with people in this way, then my advice is go buy some nice proprietary kit with a handy 1-800 number you can yell at when your expectations aren't met.
On the other hand, let's say you simply had a bad choice of words and you really do plan to participate in the crafting of the software you need
For my own use, I wandered from party to party for a long time, spending at least a few months in each while I met people, saw where things were going, got a sense of whether or not these were the sorts of people I could get along with and the sort of project where I wanted to be involved.
There was a long string of others -- I chronicalled several of them in an article/thread on advogato.org. I eventually settled at Drupal.org where the code maybe had warts (fewer all the time) and the code isn't as OO or as standards-based as maybe I'd have preferred (in other parts, like Conditional-GET, it's way too standards-based), but their archives are clean and orderly, the participatory environment is excellent (they use Drupal to build Drupal) and the core troupe are welcoming, eager, intelligent and open to new participation.
Your needs may be different, but that's enough for me.
Please consider this an innocent question, not an attack, because I truly want to know: If this method is so great and your method is so much better, why are none of these methods common in modern email or MTA software?
Actually, there's two, and both are easily found by simply entering "bayesian spam" into Google:
-
-
Please feel free to enlighten me about the above two; I'm not investing in Paul's employer, so the first issue is not nearly as important as the second, but as a spam-victim, I truly do so want to believe there's a magic anti-spam bullet, I just have trouble believing this particular story based on the data at hand.There already exists a generic method (ifile) to fold this technique into a procmail script, which means you don't need any special-purpose email program; any company that thinks its going to replace the email-browser is dreaming. I also downloaded some experimental code for Emacs GNUS, but it was too clunky for anything more than a demonstration of the rating method. Since ifile works with procmail, any ISP could use it to tag suspect email, so it wouldn't matter if there are both geeks and sex-starved teens (not the same thing?) in the audience; each can do with the extra tags as they wish.
the Google results also show this is not a new problem; research, heavy research, has been applied to Bayesian network classification of spam emails since 1994 ... so it is the first approach, or close to it. My first question is then, "if it works so well, with 0% false positives, then why did everyone, even Microsoft, abandon it?" IFile has been around for a long time, yet none of even the Linux distros include it by default. That's a little suspicious, don't you think? If the method is so foolproof, why are there no fools using it?
Excuse my greying cynicism, but there's no mention in Paul's paper of how he's accounting for the mass failure of the corpus of work that goes before him, and I get a little dubious when one lone programmer claims they can out-think large numbers of trained professionals and academics. Yes, it does happen, but when you hear hoofbeats in the street, it's usually not a zebra.
For those having trouble with the Javascript/mime-detect stuff, here's a hack tip: You will notice that the URL is an HTTP GET request where the parameter is url=http... -- that's the interface to the javascript process; if you'd like to by-pass this (and if you have Java WebStart already installed) just replace the url= with jnlp= and the Gateway will send you the JNLP file directly.
RedHat 7.1 disk one
Of course, it's only useful if several people do this all togetherRedHat 7.1 disk two
Well, for one thing, the per-block SHA hash of the trojan wouldn't match the original, so the swarm would learn to ignore that source of data. At most, you slowed down a few people imperceptably while they fetched useless blocks from your trojan server.
Not quite, but still the wrong license: He's quoting the license for the Gateway, which is the website that acts as a peer-introduction-service; the Gateway is not GPL (it's a very specific thing, not a very general tool or protocol).
Don't worry, the Gateway isn't tracking you: Your node anonymously requests a "lease" on a resource and until that lease expires, new mesh members are given your node as a possible entry-point. It's a salve until multicast discovery can do the same thing in pure peer-to-peer mode. Other apps such as Ximian, where the audience is focussed on some website, will probably also use Gateway-like software to stitch the mesh together.
But don't confuse them: There is precious little SCL code in the Gateway; it is only there to keep a temp cache of REMOTE_HOST headers and link them to known resource SHA hashed IDs.