I don't think I've ever seen Firefox grab less than 32MB of RAM, even with a blank tab:p
I've heard that VectorLinux is good for older hardare, and DamnSmallLinux (which fits on one of those miniature "business card" CDs) comes with Firefox now - might want to give those a try:)
I'm hoping so. Despite the slightly challenging tone of my post, I'm genuinely rooting for them, but so far I can think of only one automatic update that wasn't one of the minor point releases (can't think of the exact term; you know, 1.0.0, 1.0.1, 1.0.2), and these have been spread far apart indeed:(
No offense, but follow-through has been a serious problem for OSS. Look at how many projects in freshmeat are at at version 0.5 and haven't been touched in a year
That's very true, and one of the saddest sights in a the world is a promising-sounding project that has seen no releases for two years:( However, I wonder how common this would be if the authors had been given funding? I'm willing to bet that a lot of these abandoned projects were dropped mainly because the authors found that their need to put food on the table meant that they did not have the free time to carry on working on a project they loved.
Were there any speed improvements at all for day-to-day stuff not requiring transparency? I actually read the linked article this morning, and I seem to recall that there was an update at the bottom of the page with a link that expounded a little more on the relationship between XGL and Luminousity - sadly, I can't be more definite than this as the page is utterly slaughtered at the moment! In a nutshell, it was stated that Luminosity and XGL are completely separate and that neither knows much about the workings of the other, but since Luminosity acts mainly on Windows and their placement/ shaping and XGL works on the contents of the windows, the two projects should be complementary.
It's actually fairly uncanny how quickly and painlessly people make the switch from IE to Firefox. I've converted three people now, and instead of being a case of telling them to give it a few days to get used to it, and then decide whether to switch for good, all three have declared within a few minutes that they won't be going back to IE ever again. Having said that, all three were also reasonably tech-savvy, so I guess it is not as amazing as it seems, but still - you'd think the desire to stick with IE purely out of familiarity with it would be stronger.
I agree; for a browser that bills itself as being fast and light, it's collosal resource consumption is embarrasing.
Hopefully now that 1.0 has rolled around and (presumably) the majority of major features have been incorporated, more work will be done on code clean up and optimisation.
Nevertheless, that doesn't stop a misbehaving program from accidentally fork-bombing the system
I fork-bombed myself by using the Folding At Home gkrellm plug-in, once. It has an option to restart the F@H client whenever it stops, which I selected. Unfortunately, the code for detecting whether or not a F@H process was already running appeared to be faulty, and 1 minute later I had to reach for the reset switch.
It's a shame I don't have a couple of hundred processors in my desktop PC, else I would have folded the fuck out of some proteins:)
As I understand it, Apple are a *hardware* company, and make only minor profits (assuming they make any at all) on iTunes, if that's what you are referring to.
Also, if Windows users see Apple users downloading and playing only lightly-DRM'd music (which I gather is the case with iTunes music...? I've heard that Apple add it solely to appease the RIAA, although I know little of the music-downloading scene:)) whereas all of their bought media is hard-locked to their computers and could vanish as soons as a piece of hardware breaks, they might consider it an incentive to switch, no?
I'd be very surprised if Apple took the Trusted Computing route unless they absolutely had to e.g. if, as some have speculated, Web servers globally are setup not to serve non-trusted computers.
I've seen the above link and ensuing commentary many times, but I've never seen anyone critique the rebuttal above, and would be very keen to do so. Anyway, the paper above should be read as a matter of course before posting in this topic, as it is foolish to leap into a discussion having heard only one side of the argument.
Frankly, I found it allayed my fears somewhat, but I still have lingering doubts. Please reassure me:(
Apologies for the crappy URL; it seems I suck at them:)
...is the way I was introduced to Linux and the open-source world. "Open-source software never crashes!" (false); "Open-source software is nearly always better than proprietary software!" (true in some cases, but on the whole, I would say, false); "Install Linux and all your problems will be solved!" (he he he:)) These are all "facts" (albeit roughly paraphrased) that I heard, from several sources about Linux and open-source in general.
So I excitedly downloaded and burned some Mandrake 9.1 CDs, installed (amazingly painless), and entered this whole new world of Linux and Open Source.
There are many ways you could describe my reaction to the yawning gulf between expectation and reality: "furious backlash" would be one; "blind seething rage and betrayal" would be another. The catalogue of disasters and frustrations that followed left me cursing the names of the people who recommended Linux to me as this wonderful cure-all and utterly *hating* Linux with every fibre of my being. I actually stuck with it for a few more days and, after deciding that linux was the most pointless project I had ever had the misfortune of dabbling with, returned to the comparatively hassle-free bosom of Windows.
Luckily for me, the story didn't end there and some persistent and random lockups in Windows led me to re-install Linux (Mandrake 10, this time around) to see whether it was hardware or driver related (a dodgy fan on my graphics card, for what it's worth). This time around, with my expectations greatly lowered (in fact, by this stage, Linux had improved unrecognisably, and I was very pleased with its progress), I fell head-over-heels in love and haven't had Windows installed since.
But it could so easily have gone the other way; it is hard to explain just how much the hype followed by disappointment soured me on Linux at the time (an understatement!) and it is my firm belief that if I had been forewarned about the things that I could do in Windows but not in Linux, I would not have reacted so sharply. So now, when "recommending" Linux to the curious, I usually start with a list of negatives (some hardware may not work; you may not be able to run your favourite games; don't expect installation of software to follow the Windows model, or you'll be in a world of hurt) before extolling its non-pragmatic virtues ("It's an operating system with love in it!"). It's heartening that quite a few people are still willing to give it a shot even after hearing my litany of gloom:)
So in a nutshell, don't be a blind fanboy, or you will be Open Source's worst enemy; let people know that there will be concessions to make, but that many people still feel the switch is worth it. Make a special point of mentioning just how different and customisable it can be (show them a variety of WMs, from the absurdly minimalist (Ratpoision) to the wonderfully glitzy (Enlightenment) - this diversity and difference from the comparative homogeniety of Windows usually gets people curious, in my experience). Don't oversell it, as this is just a recipe for disaster.
Oh, and this post mainly deals with the casual home PC user, so might be a little off-topic - apologies for that:)
A few weeks back when I "bought" a copy of Azureus. A few weeks before that, when I "bought" a copy of hotwayd. In fact, I've spent more on OSS than I have "commercial" software! (excepting games, that is)
If someone makes a good piece of software that helps me out a lot, I like to reward their efforts. I value the "Free" aspect more than the "free".
A laptop with these features, possibly with a customised distro, would be even better (I'm not sure how much control IBM continues to exert over their ex-Laptop department, though) - this means 100% working suspend and resume with no crappy hacks required, no need to install drivers, hardware accelerated 3D + dual-head (for presentations) from the first boot, etc. Their laptops always seem to have *some* niggling incompatibility with Linux (e.g. the SD card reader on the X40) which they could probably address very easily.
I agree up to a point (for example, with my box - consisting of two different SATA drives; a TV Card; printer; scanner (which has no official XP drivers); wireless card, etc), the only thing I have to do upon a fresh install is install the nVidia binary drivers (that's assuming I want 3D acceleration - everything runs smoothly and at a good resolution with just the open source nv driver) - everything else "just works" without my having to google for, download and install drivers.
On the other hand, my cousin tried installing Linux and with a random sampling of the three network cards he had, *none* of them worked (although it is possible that simply some additional configuration was required; however, I had no idea where to start, and neither did he - ervything worked instantly under Windows). Linux is a hardware lottery, and when you win, installation is quicker and easier than Windows, but when you lose (and I'm guessing I'm in the minority in not having a single piece of hardware that is not compatible with Linux), it is a nightmare.
It's a shame no one else seemed to notice this (or saw fit to reply to it) as it sounds utterly fascinating and I would love to have had a thread dedicated to discussing what you've written. Anyway, from what you've written it sounds like you are doing great work; keep it up!:)
-- Simon
I'm late to the party, but try Klipper. Only works with text, though:( This is definitely an area where Desktop Linux is a long way behind Windows - or anything else, for that matter.
They don't give security updates to people without a licence because those people did not pay for the product
If I recall correctly, Microsoft in fact do (and will continue to) provide security updates for unlicensed installs, but not if you try to get them by visiting Windows Update; rather, it only works if Automatic Updates are turned on, which has always struck me as a better means of getting updates than manually checking, anyway. Anyone know the definite answer on this one?
I've been absolutely itching to work and this problem for months, now, but what with work on my PhD corrections, I can never find the time:( I have, however, been thinking about the problem very intensively, and the following seems to be just about tractable - if I could work on it 10 hours a day, I reckon it could be done within 8 weeks. When (if ?:( ) I start working on it, I'll be using a language like Python as coding in C/C++ will be a nightmare. Here is what I would like to (and believe I could) implement:
- "Perfect" intellisense: whenever it would make sense for a list of possible completions to be given, the program should provide them. Prerequisites are: integrated C++ Preprocessor (e.g.
typedef struct {
int a;
int b;
} A;
#define C A
C c;
c.[should output a and b as choice here]) and full knowledge of ANSI C++ rules concerning name resolution, method/ function resolution; etc. As said, this is a positively Herculean task, but given a nice, high-level, managed language (and a
copy of the ANSI C++ doc, which I've already bought:)) it is possible, given time.
- All tokens occurring in your source files tagged with meta-data. e.g. a function call should be identified with the corresponding function in your code, after resolving overloads if applicable; macros should be associated with their expansions; etc. This will have some neat side-effects, if accomplished e.g. one could search for all occasions in the code where an object of class A had a non-const method called on it, etc. This is again doable, although will be immensely heap-unfriendly:)
- All of this accomplished in real-time, including highlighting of errors.
Oddly enough, perhaps the most challenging aspect of this is to emulate the CPP preprocessor in real-time; most "tricks" that I came up with that would (I imagine) cause current would-be Intellisensers to stumble are based on doing naughty, contrived things with macros e.g.
typedef struct {
#ifdef SAUSAGE
int a;
#else
int b;
#endif
} C;
C c;
c.[is a or b that is presented as a choice?]
and even something simple like defining (perfectly validly, but very stupidly)
#define A blah }
causes a lot of problems with something simple like matching curly braces. Not to mentioned the requirement to #include source files and cache them intelligently enough to provide having to re-parse them everytime something changes (have you ever followed the chain of #inclusions through some of the standard libraries? There's a whole metric ton of shit in there!)
So this is my dream; to be able to work on this project and see it through to the stage where the real-time parsing of C++ is "perfect" to the extent that it will not fail upon encountering some of the tricks I've described above. As I mentioned, a surprisingly large amount of the problems occur from trying to apply, in real-time with a continously edited document, the rules of the preprocessor; the syntactic rules of C++ (consisting of things like how templates should be expanded, method resolution in the presence of multiple inhertitance etc), while huge and complex, are also very neatly defined in the ANSI docs, and should prove to be relatively tractable.
Perhaps an order-of-magnitude more ambitious still (although, upon recent reflection, perhaps not) would be a C++ interpreter: the rules for enacting a run of a C++ program, while still very complex, are not inordinately more complex that the rules for just parsing a C++ program. This would open up the way for one of my favourite features which has been available to Visual Studio for at least 6 years: Edit and Continue. Of course, to be useful in real life, one would have to be able to deal with linking to existing libraries, and I have no idea how this could be achieved (in that I haven't looked into it at all, not that I've looked into it and drawn a complete blank!;)) If I could writ
I agree that paper and ink is easier on the eye, but honestly, with a few days or week's practice, you can read off of a monitor screen for hours upon hours - even in the dark! Plus, you can magnify text, which helps to give one's eyes a break:)
I personally find "flipping back and forth"-type activities slightly easier with an e-book (unless you are talking about quickly ruffling the pages, looking for a specific image...?), and if the text is searchable and hyperlinked, vastly prefer it.
What's even worse are the apologists, who will go through the most amazing mental contortions to try and persuade others (and themselves) that this is not a bug or that people will be able to tell that the reported URL is fake (usually after going through some ridiculously obscure/ impractical series of steps) or that it is not at all serious. I believe that OSS is an exceptionally good model for creating secure software but please, let's not bury our heads in the sand: this is a serious problem, and "Joe Sixpack" (ugh) would not realise anything was amiss in a million years.
Slashdot Mirror
I've heard that VectorLinux is good for older hardare, and DamnSmallLinux (which fits on one of those miniature "business card" CDs) comes with Firefox now - might want to give those a try :)
I'm hoping so. Despite the slightly challenging tone of my post, I'm genuinely rooting for them, but so far I can think of only one automatic update that wasn't one of the minor point releases (can't think of the exact term; you know, 1.0.0, 1.0.1, 1.0.2), and these have been spread far apart indeed :(
Were there any speed improvements at all for day-to-day stuff not requiring transparency? I actually read the linked article this morning, and I seem to recall that there was an update at the bottom of the page with a link that expounded a little more on the relationship between XGL and Luminousity - sadly, I can't be more definite than this as the page is utterly slaughtered at the moment! In a nutshell, it was stated that Luminosity and XGL are completely separate and that neither knows much about the workings of the other, but since Luminosity acts mainly on Windows and their placement/ shaping and XGL works on the contents of the windows, the two projects should be complementary.
...and toilet paper won't be far behind...
Should go hand-in-hand with my Kommand Line Interface Transactor I'm working on :)
It's actually fairly uncanny how quickly and painlessly people make the switch from IE to Firefox. I've converted three people now, and instead of being a case of telling them to give it a few days to get used to it, and then decide whether to switch for good, all three have declared within a few minutes that they won't be going back to IE ever again. Having said that, all three were also reasonably tech-savvy, so I guess it is not as amazing as it seems, but still - you'd think the desire to stick with IE purely out of familiarity with it would be stronger.
I agree; for a browser that bills itself as being fast and light, it's collosal resource consumption is embarrasing. Hopefully now that 1.0 has rolled around and (presumably) the majority of major features have been incorporated, more work will be done on code clean up and optimisation.
It's a shame I don't have a couple of hundred processors in my desktop PC, else I would have folded the fuck out of some proteins :)
Also, if Windows users see Apple users downloading and playing only lightly-DRM'd music (which I gather is the case with iTunes music...? I've heard that Apple add it solely to appease the RIAA, although I know little of the music-downloading scene :)) whereas all of their bought media is hard-locked to their computers and could vanish as soons as a piece of hardware breaks, they might consider it an incentive to switch, no?
I'd be very surprised if Apple took the Trusted Computing route unless they absolutely had to e.g. if, as some have speculated, Web servers globally are setup not to serve non-trusted computers.
http://www.research.ibm.com/gsal/tcpa/tcpa_rebutta l.pdf
I've seen the above link and ensuing commentary many times, but I've never seen anyone critique the rebuttal above, and would be very keen to do so. Anyway, the paper above should be read as a matter of course before posting in this topic, as it is foolish to leap into a discussion having heard only one side of the argument.
Frankly, I found it allayed my fears somewhat, but I still have lingering doubts. Please reassure me :(
Apologies for the crappy URL; it seems I suck at them :)
So I excitedly downloaded and burned some Mandrake 9.1 CDs, installed (amazingly painless), and entered this whole new world of Linux and Open Source.
There are many ways you could describe my reaction to the yawning gulf between expectation and reality: "furious backlash" would be one; "blind seething rage and betrayal" would be another. The catalogue of disasters and frustrations that followed left me cursing the names of the people who recommended Linux to me as this wonderful cure-all and utterly *hating* Linux with every fibre of my being. I actually stuck with it for a few more days and, after deciding that linux was the most pointless project I had ever had the misfortune of dabbling with, returned to the comparatively hassle-free bosom of Windows.
Luckily for me, the story didn't end there and some persistent and random lockups in Windows led me to re-install Linux (Mandrake 10, this time around) to see whether it was hardware or driver related (a dodgy fan on my graphics card, for what it's worth). This time around, with my expectations greatly lowered (in fact, by this stage, Linux had improved unrecognisably, and I was very pleased with its progress), I fell head-over-heels in love and haven't had Windows installed since.
But it could so easily have gone the other way; it is hard to explain just how much the hype followed by disappointment soured me on Linux at the time (an understatement!) and it is my firm belief that if I had been forewarned about the things that I could do in Windows but not in Linux, I would not have reacted so sharply. So now, when "recommending" Linux to the curious, I usually start with a list of negatives (some hardware may not work; you may not be able to run your favourite games; don't expect installation of software to follow the Windows model, or you'll be in a world of hurt) before extolling its non-pragmatic virtues ("It's an operating system with love in it!"). It's heartening that quite a few people are still willing to give it a shot even after hearing my litany of gloom :)
So in a nutshell, don't be a blind fanboy, or you will be Open Source's worst enemy; let people know that there will be concessions to make, but that many people still feel the switch is worth it. Make a special point of mentioning just how different and customisable it can be (show them a variety of WMs, from the absurdly minimalist (Ratpoision) to the wonderfully glitzy (Enlightenment) - this diversity and difference from the comparative homogeniety of Windows usually gets people curious, in my experience). Don't oversell it, as this is just a recipe for disaster.
Oh, and this post mainly deals with the casual home PC user, so might be a little off-topic - apologies for that :)
Will be interesting to see what happens when Half Life 3 rolls along.
A few weeks back when I "bought" a copy of Azureus. A few weeks before that, when I "bought" a copy of hotwayd. In fact, I've spent more on OSS than I have "commercial" software! (excepting games, that is)
If someone makes a good piece of software that helps me out a lot, I like to reward their efforts. I value the "Free" aspect more than the "free".
The phrase "LOL" is over-used nowadays, but you got a genuine one out of me :) Bravo!
A laptop with these features, possibly with a customised distro, would be even better (I'm not sure how much control IBM continues to exert over their ex-Laptop department, though) - this means 100% working suspend and resume with no crappy hacks required, no need to install drivers, hardware accelerated 3D + dual-head (for presentations) from the first boot, etc. Their laptops always seem to have *some* niggling incompatibility with Linux (e.g. the SD card reader on the X40) which they could probably address very easily.
On the other hand, my cousin tried installing Linux and with a random sampling of the three network cards he had, *none* of them worked (although it is possible that simply some additional configuration was required; however, I had no idea where to start, and neither did he - ervything worked instantly under Windows). Linux is a hardware lottery, and when you win, installation is quicker and easier than Windows, but when you lose (and I'm guessing I'm in the minority in not having a single piece of hardware that is not compatible with Linux), it is a nightmare.
It's a shame no one else seemed to notice this (or saw fit to reply to it) as it sounds utterly fascinating and I would love to have had a thread dedicated to discussing what you've written. Anyway, from what you've written it sounds like you are doing great work; keep it up! :)
-- Simon
I'm late to the party, but try Klipper. Only works with text, though :( This is definitely an area where Desktop Linux is a long way behind Windows - or anything else, for that matter.
If a job's worth doing, etc ;) Seriously, I couldn't live with myself if I did a half-assed job of it - it's all or nothing with me, baby :)
- "Perfect" intellisense: whenever it would make sense for a list of possible completions to be given, the program should provide them. Prerequisites are: integrated C++ Preprocessor (e.g. typedef struct { int a; int b; } A; #define C A C c; c.[should output a and b as choice here]) and full knowledge of ANSI C++ rules concerning name resolution, method/ function resolution; etc. As said, this is a positively Herculean task, but given a nice, high-level, managed language (and a copy of the ANSI C++ doc, which I've already bought :)) it is possible, given time.
- All tokens occurring in your source files tagged with meta-data. e.g. a function call should be identified with the corresponding function in your code, after resolving overloads if applicable; macros should be associated with their expansions; etc. This will have some neat side-effects, if accomplished e.g. one could search for all occasions in the code where an object of class A had a non-const method called on it, etc. This is again doable, although will be immensely heap-unfriendly :)
- All of this accomplished in real-time, including highlighting of errors.
Oddly enough, perhaps the most challenging aspect of this is to emulate the CPP preprocessor in real-time; most "tricks" that I came up with that would (I imagine) cause current would-be Intellisensers to stumble are based on doing naughty, contrived things with macros e.g. typedef struct {
#ifdef SAUSAGE
int a;
#else
int b;
#endif
} C; C c; c.[is a or b that is presented as a choice?] and even something simple like defining (perfectly validly, but very stupidly)
#define A blah }
causes a lot of problems with something simple like matching curly braces. Not to mentioned the requirement to #include source files and cache them intelligently enough to provide having to re-parse them everytime something changes (have you ever followed the chain of #inclusions through some of the standard libraries? There's a whole metric ton of shit in there!)
So this is my dream; to be able to work on this project and see it through to the stage where the real-time parsing of C++ is "perfect" to the extent that it will not fail upon encountering some of the tricks I've described above. As I mentioned, a surprisingly large amount of the problems occur from trying to apply, in real-time with a continously edited document, the rules of the preprocessor; the syntactic rules of C++ (consisting of things like how templates should be expanded, method resolution in the presence of multiple inhertitance etc), while huge and complex, are also very neatly defined in the ANSI docs, and should prove to be relatively tractable.
Perhaps an order-of-magnitude more ambitious still (although, upon recent reflection, perhaps not) would be a C++ interpreter: the rules for enacting a run of a C++ program, while still very complex, are not inordinately more complex that the rules for just parsing a C++ program. This would open up the way for one of my favourite features which has been available to Visual Studio for at least 6 years: Edit and Continue. Of course, to be useful in real life, one would have to be able to deal with linking to existing libraries, and I have no idea how this could be achieved (in that I haven't looked into it at all, not that I've looked into it and drawn a complete blank! ;)) If I could writ
I personally find "flipping back and forth"-type activities slightly easier with an e-book (unless you are talking about quickly ruffling the pages, looking for a specific image...?), and if the text is searchable and hyperlinked, vastly prefer it.
What's even worse are the apologists, who will go through the most amazing mental contortions to try and persuade others (and themselves) that this is not a bug or that people will be able to tell that the reported URL is fake (usually after going through some ridiculously obscure/ impractical series of steps) or that it is not at all serious. I believe that OSS is an exceptionally good model for creating secure software but please, let's not bury our heads in the sand: this is a serious problem, and "Joe Sixpack" (ugh) would not realise anything was amiss in a million years.