In many fields, such as social sciences, most of the "code" is simply a bunch of MATLAB, Stata, or GNU R scripts, with virtually no commercial value. I'd be surprised the universities have any issues with releasing this. In fact, the faculty who want, simply post the code on their web page. The issue is that most people do not post either their code or data, and I know why. Most of the time, the code is just terrible. I know of smart tenure track faculty who rely on MATLAB to compute anything for their papers, and yet their code looks like they haven't consulted a manual for a long time. I can imagine how many published papers in fields like economics or public policy would require correction addendum once people were able to download, read and find bugs in the code.
I understand closures. In fact, you can do pretty much anything you can think of in Perl. If Python didn't exist, I guess we could still accomplish a lot with Perl.
The issue with readability and style. Look at the abundance of special characters sprinkled in the Perl code, the increased verbosity, as well as the need to understand its idiosyncratic rules, specially with respect to using references or interpreting function parameters, since Perl lacks formal named function parameters. The funny thing about Perl is that code relies on too many implicit rules (e.g. what's in @_ or $_, whether we have array or scalar context, etc) and on too many things that have to be spelled out explicitly (my for lexical scope, references for accessing nested data structure, treating function parameters explicitly as @_ array)
I like the simpler philosophy of Lisps where you can nest any data structure inside another data structure without using any references explicitly. Python follows this philosophy too. Python also uses formal function parameters, as most other languages do, but also allows them to be passed by name and without the formal order, or even omitted if there is a default value (as in GNU R). Again, you can accomplish the same tasks in Perl, but it looks too verbose, too explicit, and strangely too implicit at the same time.
I don't think anything meaningful has changed. Sure both languages have been updated. Python 3 is an improvement over Python 2. New Perl modules let you access an improved object system and features backported from Perl 6. Still, Perl remains great for writing short powerful disposable scripts in short time. Python is more suitable for writing large applications. Python is used for teaching computer science because of its dead simple syntax rules so that instructors can concentrate on teaching algorithms and concepts instead of programming language idiosyncrasies.
Awk is fine strictly as a filter. E.g. read formatted input, do something, write output. If that's all you need to do, awk is often sufficient. Perl on the other hand as a great connectivity to the operating system calls, data bases, networking, etc.
I am trying to remember where I read this, but I think I indeed heard somewhere that Python's solution under certain out of memory conditions is simply to die and exit. But still, I wouldn't discount Python as a fragile programming language. One good testament to Python's reliability is the fact that much of mailing list servers are now running GNU Mailman. GNU Mailman was what put Python on the map for me. Before Mailman, I thought no one in their right mind would write mailing list software in anything other than Perl. I was wrong.
Perl will probably linger as the language for unix scripting tasks for a long time. After all, this is how it got started. And then there is the huge library of code at CPAN that's extremely important on its own. I do think that Python is a better general purpose programming language. It's now used for a very wide range of tasks, from teaching computer science algorithms to numerical number crunching with scipy. Perl never quite got a lot of traction in such areas.
Well, the issue is that vi or rm, however dear to their users are mostly viewed as a tool, a damn good one but still a tool. But Emacs really does have a caring community around it. For example, most people in the Lisp/Scheme community use Emacs, and not by coincidence. Emacs is really a miniature programming platform for developing applications written in a version of Lisp, complete with a compiler, interpreter, debugger, and a virtual machine. Emacs's text editor itself is a Lisp application. As you can imagine, Lisp people take a special pride in this.
Next, it was already mentioned that Emacs is more than a text editor. It's almost a complete work environment. In a typical emacs session you can have multiple windows, buffers, frames open for editing multiple files, running shells, interpreters, build scripts, info sessions, etc all running at the same time.
And then don't forget the threat from Python. Python is a true general purpose scripting language. You can do web applications, sysadmin scripts, or say numerical number crunching (see scipy and numpy). It has very clean and simple syntax rules. On the other hand, Perl looks like an alien language to people who are not familiar with unix.
I agree 100%. Python sets a new standard in simplicity and readability. I don't think Perl 6 will be an improvement upon Python in this area. Python now absorbs all the new users. It even has a big traction in the CS education now (Perl never quite got there). Suffice to mention that big CS departments like MIT and Berkeley have started using Python in their intro to CS courses instead of Lisp. It's on clear to me why anyone other than Unix sysadmins should use Perl for new projects instead of Python. In fact, Python is probably just as fine as a sysadmin tool.
Perl is certainly right for sysadmins. First, Perl borrows heavily the ideas/syntax/cues from the standard unix shell scripting. I am talking about writing scripts with bash, awk, sed, grep, find, tr, etc. If you know them, you will fee right at home. Perl glues the ideas of all of these tools together into a more consistent syntax, and runs much faster than the speed most shell scripts could ever achieve.
Another important issue is the community. Perl community is filled with people who do system administration (not that there aren't other users of perl), so there are tons of libraries, which are available to use as easily as starting Perl's CPAN shell and having it install them automatically. The best book to learn Perl is Larry Walls "Programming Perl". A new edition just came out.
Having said this, I want to mention a that it's a good idea to develop a good sense of judgement. For example, I always got annoyed by some fanboyish coworkers who wrote Perl scripts when a simple shell script would suffice. I have seen perl scripts that are filled with calls to external shell commands, cp and rm and so on which I thought was stupid. (Need a shell script? just write a shell script). And I still loving using awk and similar tools for writing most of "one-liners". I always found awk to be a bit better suited for that than perl. On the other hand, know when to start writing Perl script instead of shell script. Shell scripts can get clumsy very fast.
Another advise, you may also want to check out Python. I was a Perl person, and recently looked into Python, and lo an behold, I am very impressed. In my opinion Python sets a new standard in cleanness and readability. Take a look at the free book "Dive into Python" as well as the official Python 3 tutorial online. Both are short and can be covered in just a few study sessions. Still, in sysadmin world Perl may be more useful, but Python is a great all-around general purpose language.
To see what I mean, take a look at this discussion
compare the tidy Python code at the top with the proposed Perl solutions below.
Finally, the most striking tool I have used when working as a sysadmin was CFengine. It's the bomb try it. It's a very high level declarative programming language for managing large sites/infrastructures.
This would work if companies were unable to ship the jobs overseas so easily. At least in manufacturing industries there exist at least some marginal impediments to fully free trade, but not so much in software business.
Most large universities in the US are wide open. It's a wild zoo out there. I used to work as a system administrator at a large public university, and most department managers and users were against using a central firewall. The only way around this was to configure a firewall on each individual machine.
I used to work at a university too. I was aware of security issues with printers as far back as year 2000. One shocking thing is that, not only the printer and web ports are wide open, a lot of people do not even bother to set a telnet password on them.
There are a few half baked solutions. Most printers out there have rudimentary access control capability. I have had experience with HP printers. All of them allowed me to control access by subnet number. Also, if you know that no one needs to access a printer from outside of the subnet, then leave the default gateway setting blank or 0.0.0.0. This is not perfect, but at least you know that a random web surfer from Mongolia will not stumble upon your printer's web interface.
I noticed that pretty much all new computers sold at Best Buy can be had only with Windows 8 now. So almost everyone who bought a new PC since around Thanksgiving time has Windows 8 on the new PC, and was counted as an "adopter". In reality, I know of several people who ended up buying a laptop with Windows 8 and who are horrified and want to go back. Fortunately, some vendors will mail a Windows 7 system restore CD if you request.
I have known of RMS's' distaste for C++ from as far back as 1999. Just take the KDE debacle for example. Certainly, the biggest issue that FSF had with KDE was in QT's (the library KDE was and is based on) non-free license. However, the fact that it used C++ was up there close to the license issue.
Can sympathize with them as well as with those who don't like C. I know C well, as well as several scripting languages, and just started learning C++. The complexity is just mind boggling. At the same time, using C to tackle a problem is like specifying to a carpenter where to place each molecule to build a chair. Both are very hard to code in for general problems IMO. Can't wait to move onto Java or C#.
You're setting up some kind of straw man argument here. (What can C/Java/Common Lisp/Python do that others can't do?) But I'll reply anyways.
1. CPAN libraries for the most part are excellent. Great code to use in your project or learn Perl coding from.
2. Have it occurred to you that some Perl _like_ Perl syntax? What looks unnatural to you is 100% natural code to others. Perl is great language for people who live in breath Unix. If you look at the Unix in the 80/90s, the standard way to program it was in either C or shell script (sh, csh, etc) combined with standard unix utilities like sed, awk, grep, etc. Shell scripts were ad-hockish and didn't scale well beyond a few dozen lines of code. C could do anything but was also a pain to code in, specially for simple tasks and text processing. Perl4 bridged all of these tools together in a single language. Perl5 added OOP, modules, and complex data structures.
3. Perl brought us the "Perl regular expressions", which have been widely imitated and copied by others. Perl's regular expressions are still top notch, and Perl 6 will introduce an alternative regexp system.
I am not quite sure how this idea will work in 21st century, where a lot of end products are just knowledge and technology. AMD or Intel can outsource certain chip design to a different country, no physical product actually ends being imported from there. The only result is the development of technology for the parent company, which could use it in variety of ways.
Maybe it's hard for a startup. For large and mid-sized company, it's not that hard at all to move an office abroad or outsource it completely to a 3rd party entity.
Let's set aside completely irrelevant analogies, such as your taxi drivers. We're talking about H1B visa program, which works only for people with college or advanced degrees. I can hardly imagine how the foreign H1Bs are replacing locally grown PhD. If they're worth a damn, the locals should stay on top. We have the best PhD education in the world, and this is why our graduate programs are filled +50% with foreigners.
Regarding the wages.. yes man, if you get paid $4000 a month but exactly the same job can be performed equally well by someone in Taiwan for $1000 a month, I don't see a reason for your job to exist here. However, the United States sets the pace and standards in technology. The other countries are just followers. If your job is threatened, this means it's time to update your skill or change career. This is a global economy and you can't sit still. Go get a new certification, a graduate degree or certificate, or attend a technical conference. It's not that hard to stay on top.
Build import tariffs in such a way as to severely penalize production in places where environmental and or worker safety laws are lax, make importing profits from overseas holdings difficult as well if those overseas holdings do business, directly or indirectly, in the US.
We're not talking about the outsourcing of production. We're talking about outsourcing of high skilled occupations. Think of a bunch of Indian or Russian engineers working at AMD offices in their countries or say Facebook office hiring a bunch of programmers in Indonesia. The primary purpose of these is to produce high technology knowledge for the consumption by the same company. The actual physical production, if it takes place at all, need not be done in the same county. So I don't see how your idea is going to work.
I am sorry man, but you have no idea about the reality of H1B workers. Most I have known are fairly smart people and they were already relatively well to-do by the standards of their country. They would certainly NOT come here to live 6 an apartment. Also, a lot of the successful ones eventually convert their H1B visa status to a more permanent visa to stay here. Now, it's possible that some of them send money back home. So what? Would you instead prefer to see entire corporate offices with ALL jobs moved to India, Taiwan, China, or Russia? This is not that hard at all, you know.
To put this a little blunt, this is a global competitive economy, and if you can't adapt then you should improve, change your career, or just perish. Sorry. No other way around this.
What I don't understand about many of those runaway crashes, is why no one tried to downshift the gear to let the engine braking slow down the car. Even in automatic cars, you can select 1,2,3 or at very least the "lower" gear, which is usually 2 positions below the normal. Sure, this can damage the engine, but better be safe than..
You should go check out the "base" Civic at a local dealership. For a lot of brands, the base stripper configuration is created for the sole purpose of attracting the buyer into the dealership. If you try to test drive the advertized stripper car, it will be almost certainly available in a scary color and in low number/options. Dealerships don't really like those cars. Today, no one really buys a stripper car to begin with. People want even the budget car to come with electric everything, USB/MP3/bluetooth compatible audio system, AC, reasonable interior quality, etc. For compact cars, I suspect that $20K or higher is probably the typical price people pay. In the light of this, the Prius family does not seem to be that overpriced (Honda is not really a great benchmark for hybrid cars, as Prius trumps Insight in pretty much every category)
Slashdot Mirror
In many fields, such as social sciences, most of the "code" is simply a bunch of MATLAB, Stata, or GNU R scripts, with virtually no commercial value. I'd be surprised the universities have any issues with releasing this. In fact, the faculty who want, simply post the code on their web page. The issue is that most people do not post either their code or data, and I know why. Most of the time, the code is just terrible. I know of smart tenure track faculty who rely on MATLAB to compute anything for their papers, and yet their code looks like they haven't consulted a manual for a long time. I can imagine how many published papers in fields like economics or public policy would require correction addendum once people were able to download, read and find bugs in the code.
I understand closures. In fact, you can do pretty much anything you can think of in Perl. If Python didn't exist, I guess we could still accomplish a lot with Perl.
The issue with readability and style. Look at the abundance of special characters sprinkled in the Perl code, the increased verbosity, as well as the need to understand its idiosyncratic rules, specially with respect to using references or interpreting function parameters, since Perl lacks formal named function parameters. The funny thing about Perl is that code relies on too many implicit rules (e.g. what's in @_ or $_, whether we have array or scalar context, etc) and on too many things that have to be spelled out explicitly (my for lexical scope, references for accessing nested data structure, treating function parameters explicitly as @_ array)
I like the simpler philosophy of Lisps where you can nest any data structure inside another data structure without using any references explicitly. Python follows this philosophy too. Python also uses formal function parameters, as most other languages do, but also allows them to be passed by name and without the formal order, or even omitted if there is a default value (as in GNU R). Again, you can accomplish the same tasks in Perl, but it looks too verbose, too explicit, and strangely too implicit at the same time.
I don't think anything meaningful has changed. Sure both languages have been updated. Python 3 is an improvement over Python 2. New Perl modules let you access an improved object system and features backported from Perl 6. Still, Perl remains great for writing short powerful disposable scripts in short time. Python is more suitable for writing large applications. Python is used for teaching computer science because of its dead simple syntax rules so that instructors can concentrate on teaching algorithms and concepts instead of programming language idiosyncrasies.
Indeed. It's not clear to me why bash would suddenly become that important. Please. Did I miss something?
Awk is fine strictly as a filter. E.g. read formatted input, do something, write output. If that's all you need to do, awk is often sufficient. Perl on the other hand as a great connectivity to the operating system calls, data bases, networking, etc.
I am trying to remember where I read this, but I think I indeed heard somewhere that Python's solution under certain out of memory conditions is simply to die and exit. But still, I wouldn't discount Python as a fragile programming language. One good testament to Python's reliability is the fact that much of mailing list servers are now running GNU Mailman. GNU Mailman was what put Python on the map for me. Before Mailman, I thought no one in their right mind would write mailing list software in anything other than Perl. I was wrong.
Perl will probably linger as the language for unix scripting tasks for a long time. After all, this is how it got started. And then there is the huge library of code at CPAN that's extremely important on its own. I do think that Python is a better general purpose programming language. It's now used for a very wide range of tasks, from teaching computer science algorithms to numerical number crunching with scipy. Perl never quite got a lot of traction in such areas.
Well, the issue is that vi or rm, however dear to their users are mostly viewed as a tool, a damn good one but still a tool. But Emacs really does have a caring community around it. For example, most people in the Lisp/Scheme community use Emacs, and not by coincidence. Emacs is really a miniature programming platform for developing applications written in a version of Lisp, complete with a compiler, interpreter, debugger, and a virtual machine. Emacs's text editor itself is a Lisp application. As you can imagine, Lisp people take a special pride in this.
Next, it was already mentioned that Emacs is more than a text editor. It's almost a complete work environment. In a typical emacs session you can have multiple windows, buffers, frames open for editing multiple files, running shells, interpreters, build scripts, info sessions, etc all running at the same time.
And then don't forget the threat from Python. Python is a true general purpose scripting language. You can do web applications, sysadmin scripts, or say numerical number crunching (see scipy and numpy). It has very clean and simple syntax rules. On the other hand, Perl looks like an alien language to people who are not familiar with unix.
Indeed. I prefer writing one-liners in awk. Awk remains a useful domain specific language.
And did you know that there are awk implementations that beat pretty much all other competing programming languages in speed?
See here:
http://brenocon.com/blog/2009/09/dont-mawk-awk-the-fastest-and-most-elegant-big-data-munging-language/
I agree 100%. Python sets a new standard in simplicity and readability. I don't think Perl 6 will be an improvement upon Python in this area. Python now absorbs all the new users. It even has a big traction in the CS education now (Perl never quite got there). Suffice to mention that big CS departments like MIT and Berkeley have started using Python in their intro to CS courses instead of Lisp. It's on clear to me why anyone other than Unix sysadmins should use Perl for new projects instead of Python. In fact, Python is probably just as fine as a sysadmin tool.
Perl is certainly right for sysadmins. First, Perl borrows heavily the ideas/syntax/cues from the standard unix shell scripting. I am talking about writing scripts with bash, awk, sed, grep, find, tr, etc. If you know them, you will fee right at home. Perl glues the ideas of all of these tools together into a more consistent syntax, and runs much faster than the speed most shell scripts could ever achieve.
Another important issue is the community. Perl community is filled with people who do system administration (not that there aren't other users of perl), so there are tons of libraries, which are available to use as easily as starting Perl's CPAN shell and having it install them automatically. The best book to learn Perl is Larry Walls "Programming Perl". A new edition just came out.
Having said this, I want to mention a that it's a good idea to develop a good sense of judgement. For example, I always got annoyed by some fanboyish coworkers who wrote Perl scripts when a simple shell script would suffice. I have seen perl scripts that are filled with calls to external shell commands, cp and rm and so on which I thought was stupid. (Need a shell script? just write a shell script). And I still loving using awk and similar tools for writing most of "one-liners". I always found awk to be a bit better suited for that than perl. On the other hand, know when to start writing Perl script instead of shell script. Shell scripts can get clumsy very fast.
Another advise, you may also want to check out Python. I was a Perl person, and recently looked into Python, and lo an behold, I am very impressed. In my opinion Python sets a new standard in cleanness and readability. Take a look at the free book "Dive into Python" as well as the official Python 3 tutorial online. Both are short and can be covered in just a few study sessions. Still, in sysadmin world Perl may be more useful, but Python is a great all-around general purpose language.
To see what I mean, take a look at this discussion
http://stackoverflow.com/questions/3775413/what-is-the-perl-version-of-a-python-iterator
compare the tidy Python code at the top with the proposed Perl solutions below.
Finally, the most striking tool I have used when working as a sysadmin was CFengine. It's the bomb try it. It's a very high level declarative programming language for managing large sites/infrastructures.
This would work if companies were unable to ship the jobs overseas so easily. At least in manufacturing industries there exist at least some marginal impediments to fully free trade, but not so much in software business.
Most large universities in the US are wide open. It's a wild zoo out there. I used to work as a system administrator at a large public university, and most department managers and users were against using a central firewall. The only way around this was to configure a firewall on each individual machine.
I used to work at a university too. I was aware of security issues with printers as far back as year 2000. One shocking thing is that, not only the printer and web ports are wide open, a lot of people do not even bother to set a telnet password on them.
There are a few half baked solutions. Most printers out there have rudimentary access control capability. I have had experience with HP printers. All of them allowed me to control access by subnet number. Also, if you know that no one needs to access a printer from outside of the subnet, then leave the default gateway setting blank or 0.0.0.0. This is not perfect, but at least you know that a random web surfer from Mongolia will not stumble upon your printer's web interface.
I noticed that pretty much all new computers sold at Best Buy can be had only with Windows 8 now. So almost everyone who bought a new PC since around Thanksgiving time has Windows 8 on the new PC, and was counted as an "adopter". In reality, I know of several people who ended up buying a laptop with Windows 8 and who are horrified and want to go back. Fortunately, some vendors will mail a Windows 7 system restore CD if you request.
^
One of the best posts. That's exactly my feeling too.
I have known of RMS's' distaste for C++ from as far back as 1999. Just take the KDE debacle for example. Certainly, the biggest issue that FSF had with KDE was in QT's (the library KDE was and is based on) non-free license. However, the fact that it used C++ was up there close to the license issue.
Can sympathize with them as well as with those who don't like C. I know C well, as well as several scripting languages, and just started learning C++. The complexity is just mind boggling. At the same time, using C to tackle a problem is like specifying to a carpenter where to place each molecule to build a chair. Both are very hard to code in for general problems IMO. Can't wait to move onto Java or C#.
You're setting up some kind of straw man argument here. (What can C/Java/Common Lisp/Python do that others can't do?) But I'll reply anyways.
1. CPAN libraries for the most part are excellent. Great code to use in your project or learn Perl coding from.
2. Have it occurred to you that some Perl _like_ Perl syntax? What looks unnatural to you is 100% natural code to others. Perl is great language for people who live in breath Unix. If you look at the Unix in the 80/90s, the standard way to program it was in either C or shell script (sh, csh, etc) combined with standard unix utilities like sed, awk, grep, etc. Shell scripts were ad-hockish and didn't scale well beyond a few dozen lines of code. C could do anything but was also a pain to code in, specially for simple tasks and text processing. Perl4 bridged all of these tools together in a single language. Perl5 added OOP, modules, and complex data structures.
3. Perl brought us the "Perl regular expressions", which have been widely imitated and copied by others. Perl's regular expressions are still top notch, and Perl 6 will introduce an alternative regexp system.
I am not quite sure how this idea will work in 21st century, where a lot of end products are just knowledge and technology. AMD or Intel can outsource certain chip design to a different country, no physical product actually ends being imported from there. The only result is the development of technology for the parent company, which could use it in variety of ways.
Maybe it's hard for a startup. For large and mid-sized company, it's not that hard at all to move an office abroad or outsource it completely to a 3rd party entity.
Let's set aside completely irrelevant analogies, such as your taxi drivers. We're talking about H1B visa program, which works only for people with college or advanced degrees. I can hardly imagine how the foreign H1Bs are replacing locally grown PhD. If they're worth a damn, the locals should stay on top. We have the best PhD education in the world, and this is why our graduate programs are filled +50% with foreigners.
Regarding the wages.. yes man, if you get paid $4000 a month but exactly the same job can be performed equally well by someone in Taiwan for $1000 a month, I don't see a reason for your job to exist here. However, the United States sets the pace and standards in technology. The other countries are just followers. If your job is threatened, this means it's time to update your skill or change career. This is a global economy and you can't sit still. Go get a new certification, a graduate degree or certificate, or attend a technical conference. It's not that hard to stay on top.
Build import tariffs in such a way as to severely penalize production in places where environmental and or worker safety laws are lax, make importing profits from overseas holdings difficult as well if those overseas holdings do business, directly or indirectly, in the US.
We're not talking about the outsourcing of production. We're talking about outsourcing of high skilled occupations. Think of a bunch of Indian or Russian engineers working at AMD offices in their countries or say Facebook office hiring a bunch of programmers in Indonesia. The primary purpose of these is to produce high technology knowledge for the consumption by the same company. The actual physical production, if it takes place at all, need not be done in the same county. So I don't see how your idea is going to work.
I am sorry man, but you have no idea about the reality of H1B workers. Most I have known are fairly smart people and they were already relatively well to-do by the standards of their country. They would certainly NOT come here to live 6 an apartment. Also, a lot of the successful ones eventually convert their H1B visa status to a more permanent visa to stay here. Now, it's possible that some of them send money back home. So what? Would you instead prefer to see entire corporate offices with ALL jobs moved to India, Taiwan, China, or Russia? This is not that hard at all, you know.
To put this a little blunt, this is a global competitive economy, and if you can't adapt then you should improve, change your career, or just perish. Sorry. No other way around this.
What I don't understand about many of those runaway crashes, is why no one tried to downshift the gear to let the engine braking slow down the car. Even in automatic cars, you can select 1,2,3 or at very least the "lower" gear, which is usually 2 positions below the normal. Sure, this can damage the engine, but better be safe than..
You should go check out the "base" Civic at a local dealership. For a lot of brands, the base stripper configuration is created for the sole purpose of attracting the buyer into the dealership. If you try to test drive the advertized stripper car, it will be almost certainly available in a scary color and in low number/options. Dealerships don't really like those cars. Today, no one really buys a stripper car to begin with. People want even the budget car to come with electric everything, USB/MP3/bluetooth compatible audio system, AC, reasonable interior quality, etc. For compact cars, I suspect that $20K or higher is probably the typical price people pay. In the light of this, the Prius family does not seem to be that overpriced (Honda is not really a great benchmark for hybrid cars, as Prius trumps Insight in pretty much every category)