Odd... a part of my text disappeared: However, humans don't have such a hard time with grammar, it can usually be picked up quite fast (in comparison to learning a whole new vocab and a "hard" character set). As far as these languages are concerned, it's all about vocab, and if it can pick it up and infer proper meaning from context, it could be useful. It may even work better for Korean Japanese, as I've heard those share a similar grammar.
It's not just the lack of subjects/objects in Japanese that causes a problem with (machine) translations, it's also the different sentence format. English, as well as many other European languages use the Subject-Verb-Object form, while Japanese uses the Subject-Object-Verb form. I've also seen translators having trouble with negative sentences or modal forms (can/must/must not/can not), f.e: http://translationparty.com/#10538744 Japanese, as I've heard those share a similar grammar.
Can't recall exactly, it's probably something like my slashdot username, so "gajop" or something. But seeing as that was in my freshmen/sophomore year (some 4-5 years ago), it's probably best to forget that code I produced:).
I disagree. The importance doesn't depend on how close something is to the kernel or in which software layer it resides.
It depends on what interests you have. In my case, I'm an AI researcher and a FOSS game developer of sorts (I've written, and I'm still writing AIs and tools for games). There's a notable difference between using proprietary games & engines (like Starcraft 1) to develop AIs, in comparison of using a FOSS engine (like SpringRTS).
So for some of us it really does matter, even if you don't care - in contrast I don't really care how the kernel, graphic drivers and similar low-level things are licensed, since I'll probably never want to modify it.
PS: I still more than welcome steam games for linux, mostly for the same reasons as rms, plus I won't need to restart if I play LFD2.
Is that really a good model database? It seems to have only so few models, f.e a search with a "human" tag only gives two pages, a total of 30 models and even much less complete ones. I'm not that impressed, that's as much as one open source game, I bet 0ad alone has more f.e.
I have a feeling that most games don't really post their art here, f.e. it doesn't seem like there's anything here from any of the spring games, and there's a lot of games with high quality models.
Well, even though I don't care how you Americans spend money, I can agree that it's unlikely to bear any fruit any time soon.
However, if it did somehow manage to detect and decipher alien communication, it would be huge. It would change beliefs, and in case of a technologically advanced civilization, it would even make discovery and understanding of alien tech nearly as important if not more than an actual direct pursuit of science - why do it yourself if you can see how someone else did it, as most of the fundamental science would probably be the same.
Biggest problem I had with awesome (some 2-3 years ago) were the big changes that would brake any custom config/widgets I wrote. The main thing was that switch to Lua, but it didn't stop there, I would end up spending an hour or so after a big change - even when running it in xephyr to help with debugging the new errors.
Personally, I use Openbox and trays/panels that haven't changed much and still compile fine for the last 4 years. If i need tiling I just use tmux.
What puzzles me is that these online courses don't exploit two main features: they're all digital and there's very little interaction with the teaching stuff that's necessary. Why do they need to be repeated? Why can't I just sign up and do entire courses at my own pace? At least don't restrict the start period.
F.e. coursera had an interesting course about probabilistic graph models, but I didn't notice it until it ended, so I couldn't enroll to do it entirely, and could only watch the videos, without being able to work on any exams.
Recently I enrolled into Thrun's 2nd Robotic Car course, but I had to postpone it until I was done with my masters. The first few lessons I did ahead of schedule and to me it didn't seem there was any work done to modify the original course - I recall repeatedly checking the "Office week" videos to see if any of our new questions were answered, without any changes.
So please, don't re-run courses. If you really do have something to add, just modify a couple of videos/problems and notify students that you're doing a "second release", rather than a new semester - that old teaching model doesn't need to be applied here.
Regarding package management, as far as the users are concerned I'd say Arch has the best possible thing. I've tested portage(Gentoo), apt-get/aptitude (Debian&Ubuntu), fedora's rpm installs (can't remember the apps name). Speed: Package manager (pacman) is probably the fastest one there is, when testing repository querying, installs, local queries and so on. Some operations got even faster with the new install. Packages: A lot of packages are available on core/extra/community repositories, most of what an average user would need. Packages exist for both x86 and x86_64 as well as those for cross-compilations in multilib. The few that aren't there will probably be found in AUR. Package customization: You can create your own PKGBUILDS, simple files (much simpler than what you'd have to do in Gentoo f.e) that do package compilation and installation. You can even take existing ones and modify them ever so slightly to fit your needs:every package in core/extra/community has PKGBUILDS downloadable with simple commands (f.e "abs extra/firefox").
I personally don't care about package signing, if I'm ever concerned about a package (from AUR only, core/extra/community I trust more than official sites/sources), I could always take a fast skim at PKGBUILD and make sure it doesn't do anything weird (the logic is just a bunch of sh commands after all).
Oh and I forgot to mention one thing: Arch Wiki - probably making it the best documented distro there is, and #archlinux on freenode, where you'll usually always get information on what you need (it's even usable as an information source for non-arch related things).
Read: https://wiki.archlinux.org/index.php/Arch_Compared_to_Other_Distributions I don't think you have a clue tbh. I've tried most well known Linuxes (all that you mentioned and a few others), and I can tell you that there are two major differences that distros have, as far as users are concerned: 1) GUI/CLI based (which is also complex/minimalistic), 2) Regular/rolling release based.
1) Ubuntu, Fedora, OpenSUSE and so on are GUI based systems, coming with fully installed DEs and offering people little choice on the initial install. Sure you can remove stuff and install simple WMs, but that just makes it harder to configure than Arch/Gentoo and even Slackware, who are made for ground-up installation. The reason I use Arch regularly is because I can configure it to do pretty much exactly what I want.
2) Ubuntu, Fedora, Debian, OpenSUSE, Slackware, and a whole lot of others are using the regular (once, twice a year) release cycle. It's fine if you're using it in the office/classroom/servers, or you just don't use computers much. But often, software updates come a lot more regularly than that (Windows _software_ is rolling release!, the OS itself isn't of course), and it's always good to in the bleeding edge - unless it's you who's bleeding, and that's a potential problem (much like this update required some meddling before it would just work). And even if you do get problems every once in a while when you do rolling release updates, the huge amount of problems whenever I do a full update every 6 month on Ubuntu makes me want to do a clean install (I'm using an uptodate Arch from 2008~, did some experimenting with other linuxes). In the rolling release field it's quite similar to Gentoo (that's another power of Gentoo, it isn't just people compiling stuff for the laughs).
As long as the ratio is maintained I'm ok with it. I've seen many books labelled "Learn how to program in {24, 7} days.", however I've also seen the book "Learn how to become a manager in 1 minute" so I think all is as it should be.
From the video, they grew their food in what appears to be elevator space between two levels. Is this really enough to feed two adults? I always assumed you'd need at least another large room, perhaps even with specialized equipment such as growing lamps to be able to grow it 24/7, also being able to utilize volume (by having multiple levels), not just the surface.
And what are these services? All you've done is say multiple times how they're different, you even included a car analogy, but you failed to name a single service. Sorry, that doesn't work for me.
*DEs are just highly bloated WMs where all the choices have been made for you, but there's no reason your WM can't be as powerful as any *DE (it often is more).
That's all well in theory, but whenever I try to use one of those "Great User Experience" distros things just fall apart eventually. They nearly always have bugs in certain essential system parts, like graphic drivers breaking (happened today on Linux Mint so it's rather fresh), printer problems (still can't make HP1020 work well in Ubuntu), proper home (static IP, routing, ethernet + wifi depending on connection) networking (was a problem when I last checked Ubuntu, hopefully fixed by now), working utilities (music, cd/dvd burning all broke on Ubuntu for me), working options (today, Linux Mint asked if I wanted to copy pidgin settings from Ubuntu, I said yes, it didn't work, my "cp $TARGET $DEST" did).
It even gets worse when you do system updates such as 10.04->10.10 for Ubuntu, on systems that don't have a rolling release model. Then you will get a bunch of more new problems, many which will not get fixed by subsequent software updates but require settings fiddling or a new install. Sure, I too tend to work on higher level (abstraction) programming tasks (best done by matlab like programs) mostly, and I'd love to be able to get rid of all the configuration fiddling when I'm not setting my own preferences, but it's just not possible now (when distros just announce new GUI toolkits instead of making the system more stable - Windows 7 is more stable than any GUI Linux I ever used, and I used a lot of them).
The answer is clear, if you want a complete "build it yourself" distribution, with parts hand choosen, just go for one of the command line interface based distributions, such as Arch or Gentoo, which come with a bare system. F.e by just following Arch Linux' wiki for system installation you will get familiar with all the WM/DE choices, and depending on what you pick there you can get further specific information on the Arch wiki or specific WM, regarding systray/pager/filemanager and other utilities that work well there.
I for one have openbox with tint2, conky and pypanel, with thunar as filemanager (although I often just use coreutils when it's faster/easier). Of course, no one is forcing you to choose Arch or Gentoo, Ubuntu is fine but to me it makes no sense to choose a GUI distribution which comes hand polished for GNOME/KDE/*DE usage when you will just clean it all and install ratpoison.
Hmm, similar article so I'll ask a question of personal nature.
I've recently created a crawler to collect certain information from a website, that would help me gather data sets for a small machine learning project. While I've followed robots.txt and nofollow links, site's TOU was against it. After confirming with the admin, I was told that it's not allowed to gather information, as the site owns it (as it's written in the TOU).
The data however is publicly available, so you actually wouldn't have to agree to a TOU to collect the data, and as it's some data I wanted, I still concluded I should get a small sample (less than 1% of the total data, around 200MB) at least, to see if something's even possible to be done with it.
What are your thoughts/.? Should I have abandoned the attempt, have I done right or even should I disregard their plead and simply get as much as I please (during a long period of time, as to not hammer on it's bandwidth)?
Knowing full well much of/. will be up in pitchforks stating how they're out to get you, I think it's a good idea. A professor of mine worked on a similar project (competition), creating data mining (machine learning) software that predicts whether someone has suicidal tendencies.
Just remember, most people working with data mining aren't doing it because they are interested in your personal life, but rather wish to help people and companies have better lives and more money.
This has nothing to do with whether medical care should be funded by you directly, or someone else (the government). And I don't quite get why you assume that research in medicine (or any other that field for that matter) should by default become public property, it's definitely not cheap.
This is a result of hard work and ingenuity, and unless you reward that, by allowing people to make a living through research, you will never incentivize people to do it. I would much rather have millionaires and billionaires coming from research rather than management of world fast food restaurants.
I'd rather promote people funding research knowing full well they can make money out of it, if it means curing many of the world diseases.
Even if it was, it's ok if it'd be kept behind patents and intellectual property law as long as they'd sell it, and I bet they would. Or do you think 100k is what you get when you curse cancer?
Slashdot Mirror
Odd... a part of my text disappeared:
However, humans don't have such a hard time with grammar, it can usually be picked up quite fast (in comparison to learning a whole new vocab and a "hard" character set). As far as these languages are concerned, it's all about vocab, and if it can pick it up and infer proper meaning from context, it could be useful. It may even work better for Korean Japanese, as I've heard those share a similar grammar.
It's not just the lack of subjects/objects in Japanese that causes a problem with (machine) translations, it's also the different sentence format.
English, as well as many other European languages use the Subject-Verb-Object form, while Japanese uses the Subject-Object-Verb form.
I've also seen translators having trouble with negative sentences or modal forms (can/must/must not/can not), f.e: http://translationparty.com/#10538744 Japanese, as I've heard those share a similar grammar.
Can't recall exactly, it's probably something like my slashdot username, so "gajop" or something. But seeing as that was in my freshmen/sophomore year (some 4-5 years ago), it's probably best to forget that code I produced :).
Ages ago I contributed a minor patch or two for that project, glad to hear you guys are still alive and kicking :D
Or you can just use Spring: http://springrts.com/, which is FOSS, cross platform, and has a bunch of working games already: http://springrts.com/wiki/Games.
I disagree.
The importance doesn't depend on how close something is to the kernel or in which software layer it resides.
It depends on what interests you have.
In my case, I'm an AI researcher and a FOSS game developer of sorts (I've written, and I'm still writing AIs and tools for games).
There's a notable difference between using proprietary games & engines (like Starcraft 1) to develop AIs, in comparison of using a FOSS engine (like SpringRTS).
So for some of us it really does matter, even if you don't care - in contrast I don't really care how the kernel, graphic drivers and similar low-level things are licensed, since I'll probably never want to modify it.
PS: I still more than welcome steam games for linux, mostly for the same reasons as rms, plus I won't need to restart if I play LFD2.
Is that really a good model database? It seems to have only so few models, f.e a search with a "human" tag only gives two pages, a total of 30 models and even much less complete ones.
I'm not that impressed, that's as much as one open source game, I bet 0ad alone has more f.e.
I have a feeling that most games don't really post their art here, f.e. it doesn't seem like there's anything here from any of the spring games, and there's a lot of games with high quality models.
Well, even though I don't care how you Americans spend money, I can agree that it's unlikely to bear any fruit any time soon.
However, if it did somehow manage to detect and decipher alien communication, it would be huge.
It would change beliefs, and in case of a technologically advanced civilization, it would even make discovery and understanding of alien tech nearly as important if not more than an actual direct pursuit of science - why do it yourself if you can see how someone else did it, as most of the fundamental science would probably be the same.
Biggest problem I had with awesome (some 2-3 years ago) were the big changes that would brake any custom config/widgets I wrote.
The main thing was that switch to Lua, but it didn't stop there, I would end up spending an hour or so after a big change - even when running it in xephyr to help with debugging the new errors.
Personally, I use Openbox and trays/panels that haven't changed much and still compile fine for the last 4 years. If i need tiling I just use tmux.
there are some common dumps, like http://wiki.dbpedia.org/Downloads37
What puzzles me is that these online courses don't exploit two main features: they're all digital and there's very little interaction with the teaching stuff that's necessary. Why do they need to be repeated? Why can't I just sign up and do entire courses at my own pace? At least don't restrict the start period.
F.e. coursera had an interesting course about probabilistic graph models, but I didn't notice it until it ended, so I couldn't enroll to do it entirely, and could only watch the videos, without being able to work on any exams.
Recently I enrolled into Thrun's 2nd Robotic Car course, but I had to postpone it until I was done with my masters. The first few lessons I did ahead of schedule and to me it didn't seem there was any work done to modify the original course - I recall repeatedly checking the "Office week" videos to see if any of our new questions were answered, without any changes.
So please, don't re-run courses. If you really do have something to add, just modify a couple of videos/problems and notify students that you're doing a "second release", rather than a new semester - that old teaching model doesn't need to be applied here.
Well, we routinely blackout a smaller site of choice, just post an article about it.
)
I agree.
Regarding package management, as far as the users are concerned I'd say Arch has the best possible thing. I've tested portage(Gentoo), apt-get/aptitude (Debian&Ubuntu), fedora's rpm installs (can't remember the apps name). :every package in core/extra/community has PKGBUILDS downloadable with simple commands (f.e "abs extra/firefox").
Speed: Package manager (pacman) is probably the fastest one there is, when testing repository querying, installs, local queries and so on. Some operations got even faster with the new install.
Packages: A lot of packages are available on core/extra/community repositories, most of what an average user would need. Packages exist for both x86 and x86_64 as well as those for cross-compilations in multilib. The few that aren't there will probably be found in AUR.
Package customization: You can create your own PKGBUILDS, simple files (much simpler than what you'd have to do in Gentoo f.e) that do package compilation and installation. You can even take existing ones and modify them ever so slightly to fit your needs
I personally don't care about package signing, if I'm ever concerned about a package (from AUR only, core/extra/community I trust more than official sites/sources), I could always take a fast skim at PKGBUILD and make sure it doesn't do anything weird (the logic is just a bunch of sh commands after all).
Oh and I forgot to mention one thing: Arch Wiki - probably making it the best documented distro there is, and #archlinux on freenode, where you'll usually always get information on what you need (it's even usable as an information source for non-arch related things).
Read: https://wiki.archlinux.org/index.php/Arch_Compared_to_Other_Distributions
I don't think you have a clue tbh. I've tried most well known Linuxes (all that you mentioned and a few others), and I can tell you that there are two major differences that distros have, as far as users are concerned: 1) GUI/CLI based (which is also complex/minimalistic), 2) Regular/rolling release based.
1) Ubuntu, Fedora, OpenSUSE and so on are GUI based systems, coming with fully installed DEs and offering people little choice on the initial install. Sure you can remove stuff and install simple WMs, but that just makes it harder to configure than Arch/Gentoo and even Slackware, who are made for ground-up installation. The reason I use Arch regularly is because I can configure it to do pretty much exactly what I want.
2) Ubuntu, Fedora, Debian, OpenSUSE, Slackware, and a whole lot of others are using the regular (once, twice a year) release cycle. It's fine if you're using it in the office/classroom/servers, or you just don't use computers much. But often, software updates come a lot more regularly than that (Windows _software_ is rolling release!, the OS itself isn't of course), and it's always good to in the bleeding edge - unless it's you who's bleeding, and that's a potential problem (much like this update required some meddling before it would just work). And even if you do get problems every once in a while when you do rolling release updates, the huge amount of problems whenever I do a full update every 6 month on Ubuntu makes me want to do a clean install (I'm using an uptodate Arch from 2008~, did some experimenting with other linuxes). In the rolling release field it's quite similar to Gentoo (that's another power of Gentoo, it isn't just people compiling stuff for the laughs).
As long as the ratio is maintained I'm ok with it. I've seen many books labelled "Learn how to program in {24, 7} days.", however I've also seen the book "Learn how to become a manager in 1 minute" so I think all is as it should be.
From the video, they grew their food in what appears to be elevator space between two levels.
Is this really enough to feed two adults? I always assumed you'd need at least another large room, perhaps even with specialized equipment such as growing lamps to be able to grow it 24/7, also being able to utilize volume (by having multiple levels), not just the surface.
And what are these services? All you've done is say multiple times how they're different, you even included a car analogy, but you failed to name a single service. Sorry, that doesn't work for me.
*DEs are just highly bloated WMs where all the choices have been made for you, but there's no reason your WM can't be as powerful as any *DE (it often is more).
That's all well in theory, but whenever I try to use one of those "Great User Experience" distros things just fall apart eventually.
They nearly always have bugs in certain essential system parts, like graphic drivers breaking (happened today on Linux Mint so it's rather fresh), printer problems (still can't make HP1020 work well in Ubuntu), proper home (static IP, routing, ethernet + wifi depending on connection) networking (was a problem when I last checked Ubuntu, hopefully fixed by now), working utilities (music, cd/dvd burning all broke on Ubuntu for me), working options (today, Linux Mint asked if I wanted to copy pidgin settings from Ubuntu, I said yes, it didn't work, my "cp $TARGET $DEST" did).
It even gets worse when you do system updates such as 10.04->10.10 for Ubuntu, on systems that don't have a rolling release model. Then you will get a bunch of more new problems, many which will not get fixed by subsequent software updates but require settings fiddling or a new install.
Sure, I too tend to work on higher level (abstraction) programming tasks (best done by matlab like programs) mostly, and I'd love to be able to get rid of all the configuration fiddling when I'm not setting my own preferences, but it's just not possible now (when distros just announce new GUI toolkits instead of making the system more stable - Windows 7 is more stable than any GUI Linux I ever used, and I used a lot of them).
TL:DR https://bbs.archlinux.org/viewtopic.php?id=131196 read the information below the screenshots and take your pick! Your realization is what people were doing for many years now.
The answer is clear, if you want a complete "build it yourself" distribution, with parts hand choosen, just go for one of the command line interface based distributions, such as Arch or Gentoo, which come with a bare system.
F.e by just following Arch Linux' wiki for system installation you will get familiar with all the WM/DE choices, and depending on what you pick there you can get further specific information on the Arch wiki or specific WM, regarding systray/pager/filemanager and other utilities that work well there.
I for one have openbox with tint2, conky and pypanel, with thunar as filemanager (although I often just use coreutils when it's faster/easier). Of course, no one is forcing you to choose Arch or Gentoo, Ubuntu is fine but to me it makes no sense to choose a GUI distribution which comes hand polished for GNOME/KDE/*DE usage when you will just clean it all and install ratpoison.
Hmm, similar article so I'll ask a question of personal nature.
I've recently created a crawler to collect certain information from a website, that would help me gather data sets for a small machine learning project.
While I've followed robots.txt and nofollow links, site's TOU was against it. After confirming with the admin, I was told that it's not allowed to gather information, as the site owns it (as it's written in the TOU).
The data however is publicly available, so you actually wouldn't have to agree to a TOU to collect the data, and as it's some data I wanted, I still concluded I should get a small sample (less than 1% of the total data, around 200MB) at least, to see if something's even possible to be done with it.
What are your thoughts /.? Should I have abandoned the attempt, have I done right or even should I disregard their plead and simply get as much as I please (during a long period of time, as to not hammer on it's bandwidth)?
Knowing full well much of /. will be up in pitchforks stating how they're out to get you, I think it's a good idea.
A professor of mine worked on a similar project (competition), creating data mining (machine learning) software that predicts whether someone has suicidal tendencies.
Just remember, most people working with data mining aren't doing it because they are interested in your personal life, but rather wish to help people and companies have better lives and more money.
This has nothing to do with whether medical care should be funded by you directly, or someone else (the government).
And I don't quite get why you assume that research in medicine (or any other that field for that matter) should by default become public property, it's definitely not cheap.
This is a result of hard work and ingenuity, and unless you reward that, by allowing people to make a living through research, you will never incentivize people to do it.
I would much rather have millionaires and billionaires coming from research rather than management of world fast food restaurants.
I'd rather promote people funding research knowing full well they can make money out of it, if it means curing many of the world diseases.
Even if it was, it's ok if it'd be kept behind patents and intellectual property law as long as they'd sell it, and I bet they would.
Or do you think 100k is what you get when you curse cancer?
If anything, these robots will be used to make shivs. Well, at least it's not a bipod.