Ah, yes, working from home. As I am doing right now. Unfortunately, I have to admit that you can communicate far more efficiently when you are *in* the environment than from your home office. But perhaps technology will in time solve that issue.
I think you just take the research grants where they are to be found. I mean, I have done optimization work too, where the real agenda was to save on fuel costs, but the research was officially "green":)
Well, most families have 2 working people (at least in the civilized world), which means that usually one of them have half an hour drive time or more (for me, around 30 km). With public transportation, that is about 50 minutes, which does not make me "uncontrollably angry", but I wouldn't mind seeing it reduced, either.
It demonstrates nothing, really, and if it did it was consistency rather than flexibility. Linux is not exactly known for consistency, though obviously whoever did the shortcuts for the shell was an emacs user. You could say the flexibility is that you can customize these keys, but that is true of most shells, as far as I am aware.
Frankly, I do not think most people can appreciate the power in Linux immediately. Hence, I vote for sowing the seed (the warm-fuzzy language as you call it), preferably backed up with real-life use cases. And I vote for teaching bare survival, for which the arrow keys, insert, del and backspace is more than enough for editing at a first go. And sure, a few key shell commands.... or maybe just hand out a scream sheet and some exercises. I do not vote for hard-to-learn editors (vi, emacs are both in this category though vi is worse) --- in fact, for computer science students I'd jump directly to some appropriate IDE for whatever language they will use (KDevelop, Eclipse, etc.), plus whatever supporting tools they will have to know. After the initial shock, it is soo much easier to expand.
What can I say? I disagree. The power of Linux is not about ctrl-t to swap 2 chars in a command line. It's the openness, the flexibility and the power of the legion.
You need to imagine your DB needs to be so big that storing it on a dozen computers is not enough and imagine this DB is updated as well as read. From this follows that you need more or less transparent redundancy, since some of those servers will be failing at a given time. Oh, and you need to be able to add new servers as needed. And do it on cheapish hardware.
It is a scenario outside of my experience, and, I would guess, outside yours.
These days, the arrow keys works almost everywhere as you would expect, as does del, backspace and insert and typing That's enough to last the first few years.
The only exception is vi. For vi, the most important command to learn for vi is how to get out, as none of the conventional shortcuts works for vi.
Seriously, vi(m) is an editor. Some like it, some do not, but it is not anything but an editor. Editors are dime a dozen, and most of them can do as much as vi can.
Compile times is the bottom of the pack for me. Modern computers are so fast, I seldom sit and wait for compilation. C++ already offers peak performance.
Still, I might have a look again. Last time around I was not impressed:)
LOL. Mathematicians tend to be poor at building abstractions?! ROFL Pull the other one, it has bells on!
Sure, it would be nice if C++ supported duck typing, but such can be emulated with simple encapsulation and it sort of works. If that was what you were referring to.
Sure, and nobody has lost anything compared to if the site was not there. People giving up that easily are not worth anything to an OS project in any case.
C# is a white spot I admit. It's on my todo list, but what I hear from people who use it is "meh. Better than Java, but far from C++".
Java is hardly an improvement to C++ by any conventional definition, as it lacks a ton of features that C++ sports. It does have a bigger marketing department, and that shows.
I haven't tried OCaml, but I have tried Haskell. Cute for toy projects, but again, lacking a lot of features I need to be as effective as I am in C++. In particular, I found it too bothersome to do OO programming, which have me a lot of problems when trying to tackle non-trivial problems. It's been a while, though, and back then those features were supposedly in the pipeline, so maybe it is time to try again.
It is not uncommon for a new language to be somewhat compatible with another language. In fact, C++ itself is a prime example of this.
That said, you have a strange idea of "didn't work out". Namespaces are a critical feature to C++, [] is used for lists today (and indeed for the very std::string class you mention). While the library interfaces could always stand some improvement, I hardly find it a showstopper that a few interfaces are still on char* --- there might even be good reasons for it (think about it).
OK. So you were burned.. 10 years ago? Maybe it is time to move on? I have not seen a self-written Vector or String class for many years now, excepting the Qt-ones. And thanks to the flexibility of the C++ language, those Qt-classes integrate pretty well with the std ones.
I do agree, though, that C++ sucks for a number of reasons. E.g. default fall-through in case statements is moronic. I only claim that it is the language I keep returning to because while it sucks, it sucks less than everything else I have tried and it has more and better libraries for most things.
Well, don't toy with me like that! What language do you suggest? Note I have tried quite a few...
You got your timeline screwed up, btw. Declare-everything didn't hit mainstream until long after the 1960's. Still old, sure, but new isn't necessarily better.
You know, I did write (and do, actually) unit tests --- and I get paid, so definitely professionally. My rubv code is still in production, virtually unchanged, so it was not THAT bad. And no, that did not help me enough, though the test features of ruby are excellent. As ruby programs grow, I spent too long hunting for hidden type errors. It's not that I could not do it, it's that I am vastly more productive in C++. Even though C++ writes function declarations backwards, which is hugely annoying.
Note that maybe it is a background thing. I am a mathematician by profession, so working with types (well, sets really) comes very naturally to me, and I am seldom limited by the type system in any way... on the contrary, it helps me see the structure as it unfolds itself.
Java sucks. It lacks a lot of essential features: Operator overloading, static polymorphism, lambdas/closures, multiple inheritance (or at LEAST mix-ins), const-support, proper support for limited resources (besides memory)... I could go on. Add to that, the JVMs I see today handle memory extremely poorly, meaning I and many other tend to avoid Java programs whenever possible. Added to that comes the current patent problems.
Java is a suitable replacement for COBOL, and little more. As for speed... Eclipse is sluggish, KDevelop/Qt-creator is fast --- need I say more? Not that that in itself would bother me for most tasks.
And don't get me started on JNI. It's cumbersome, slow, buggy and most of all ONE WAY! How fucked up is that?
Sorry, I had to work in Java again recently, I had almost regressed what a straightjacket of a language that is. Though at least the C#-competition forced a few features that were long overdue..anyone remember how it was to work with the old Collections? "Please leave your type information at entry, or create a freaking huge boilerplate class for every type you need to use".
I never understood the first complaint. Yes, you can make = or + or whatever do crazy thing you like, but that goes for every function call. I remember debugging some (Java code, but that is besides the point) where the programmer had decided that getFoo() should actually change the internal state of the object in such a way that calling getFoo() twice totally screwed up the program. That was every bit as bad as if he has overloaded operator- to do the same.
Slicing, I agree, is a real problem, though I have seldom encountered it in real life. Perhaps because I like to keep all but my leaf classes abstract, which more or less eliminates the slicing problem.
This will be the, what, second standard? Third? And besides the new libraries, it's not that many new hard-to-grok features that are added. Actually, I can only think of lambda, but a language without lambda is an atrocity. Is there any particular feature you are thinking of? Have you read the wikipedia page?
I have worked on a number of projects, usually as some sort of project lead or technical lead. I have had problems with people who could not grok C++ (though not many), but I have not had the problem you describe, which just shows how lucky I am;)
In my humble opinion the biggest problem I have had over the years is to get people to search for a freaking library instead of trying to do everything themselves. Admittedly, that is not a problem specific to C++.
Slashdot Mirror
Ah, yes, working from home. As I am doing right now. Unfortunately, I have to admit that you can communicate far more efficiently when you are *in* the environment than from your home office. But perhaps technology will in time solve that issue.
I think you just take the research grants where they are to be found. I mean, I have done optimization work too, where the real agenda was to save on fuel costs, but the research was officially "green" :)
Well, most families have 2 working people (at least in the civilized world), which means that usually one of them have half an hour drive time or more (for me, around 30 km). With public transportation, that is about 50 minutes, which does not make me "uncontrollably angry", but I wouldn't mind seeing it reduced, either.
for the very rich, if it takes much Pa
It demonstrates nothing, really, and if it did it was consistency rather than flexibility. Linux is not exactly known for consistency, though obviously whoever did the shortcuts for the shell was an emacs user. You could say the flexibility is that you can customize these keys, but that is true of most shells, as far as I am aware.
Frankly, I do not think most people can appreciate the power in Linux immediately. Hence, I vote for sowing the seed (the warm-fuzzy language as you call it), preferably backed up with real-life use cases. And I vote for teaching bare survival, for which the arrow keys, insert, del and backspace is more than enough for editing at a first go. And sure, a few key shell commands.... or maybe just hand out a scream sheet and some exercises. I do not vote for hard-to-learn editors (vi, emacs are both in this category though vi is worse) --- in fact, for computer science students I'd jump directly to some appropriate IDE for whatever language they will use (KDevelop, Eclipse, etc.), plus whatever supporting tools they will have to know. After the initial shock, it is soo much easier to expand.
What can I say? I disagree. The power of Linux is not about ctrl-t to swap 2 chars in a command line. It's the openness, the flexibility and the power of the legion.
You need to imagine your DB needs to be so big that storing it on a dozen computers is not enough and imagine this DB is updated as well as read. From this follows that you need more or less transparent redundancy, since some of those servers will be failing at a given time. Oh, and you need to be able to add new servers as needed. And do it on cheapish hardware.
It is a scenario outside of my experience, and, I would guess, outside yours.
These days, the arrow keys works almost everywhere as you would expect, as does del, backspace and insert and typing That's enough to last the first few years.
The only exception is vi. For vi, the most important command to learn for vi is how to get out, as none of the conventional shortcuts works for vi.
Don't forget to teach them punchcards, too!
Seriously, vi(m) is an editor. Some like it, some do not, but it is not anything but an editor. Editors are dime a dozen, and most of them can do as much as vi can.
Yeah, yeah, I see you sig. Well done.
Compile times is the bottom of the pack for me. Modern computers are so fast, I seldom sit and wait for compilation. C++ already offers peak performance.
Still, I might have a look again. Last time around I was not impressed :)
LOL. Mathematicians tend to be poor at building abstractions?! ROFL Pull the other one, it has bells on!
Sure, it would be nice if C++ supported duck typing, but such can be emulated with simple encapsulation and it sort of works. If that was what you were referring to.
Sure, and nobody has lost anything compared to if the site was not there. People giving up that easily are not worth anything to an OS project in any case.
C# is a white spot I admit. It's on my todo list, but what I hear from people who use it is "meh. Better than Java, but far from C++".
Java is hardly an improvement to C++ by any conventional definition, as it lacks a ton of features that C++ sports. It does have a bigger marketing department, and that shows.
I haven't tried OCaml, but I have tried Haskell. Cute for toy projects, but again, lacking a lot of features I need to be as effective as I am in C++. In particular, I found it too bothersome to do OO programming, which have me a lot of problems when trying to tackle non-trivial problems. It's been a while, though, and back then those features were supposedly in the pipeline, so maybe it is time to try again.
It is not uncommon for a new language to be somewhat compatible with another language. In fact, C++ itself is a prime example of this.
That said, you have a strange idea of "didn't work out". Namespaces are a critical feature to C++, [] is used for lists today (and indeed for the very std::string class you mention). While the library interfaces could always stand some improvement, I hardly find it a showstopper that a few interfaces are still on char* --- there might even be good reasons for it (think about it).
Send a patch, then :)
OK. So you were burned.. 10 years ago? Maybe it is time to move on? I have not seen a self-written Vector or String class for many years now, excepting the Qt-ones. And thanks to the flexibility of the C++ language, those Qt-classes integrate pretty well with the std ones.
I do agree, though, that C++ sucks for a number of reasons. E.g. default fall-through in case statements is moronic. I only claim that it is the language I keep returning to because while it sucks, it sucks less than everything else I have tried and it has more and better libraries for most things.
Yeah, yeah, English is my second language and I was tired. SORRY!! :)
Well, don't toy with me like that! What language do you suggest? Note I have tried quite a few...
You got your timeline screwed up, btw. Declare-everything didn't hit mainstream until long after the 1960's. Still old, sure, but new isn't necessarily better.
You know, I did write (and do, actually) unit tests --- and I get paid, so definitely professionally. My rubv code is still in production, virtually unchanged, so it was not THAT bad. And no, that did not help me enough, though the test features of ruby are excellent. As ruby programs grow, I spent too long hunting for hidden type errors. It's not that I could not do it, it's that I am vastly more productive in C++. Even though C++ writes function declarations backwards, which is hugely annoying.
Note that maybe it is a background thing. I am a mathematician by profession, so working with types (well, sets really) comes very naturally to me, and I am seldom limited by the type system in any way... on the contrary, it helps me see the structure as it unfolds itself.
Java sucks. It lacks a lot of essential features: Operator overloading, static polymorphism, lambdas/closures, multiple inheritance (or at LEAST mix-ins), const-support, proper support for limited resources (besides memory)... I could go on. Add to that, the JVMs I see today handle memory extremely poorly, meaning I and many other tend to avoid Java programs whenever possible. Added to that comes the current patent problems.
Java is a suitable replacement for COBOL, and little more. As for speed... Eclipse is sluggish, KDevelop/Qt-creator is fast --- need I say more? Not that that in itself would bother me for most tasks.
And don't get me started on JNI. It's cumbersome, slow, buggy and most of all ONE WAY! How fucked up is that?
Sorry, I had to work in Java again recently, I had almost regressed what a straightjacket of a language that is. Though at least the C#-competition forced a few features that were long overdue..anyone remember how it was to work with the old Collections? "Please leave your type information at entry, or create a freaking huge boilerplate class for every type you need to use".
Those ARE atrocious, are they not? Personally, my coding standing are usually "types looks like this, member variables like this, code nice".
Of course, the last bit takes a bit for people to learn, but it works nicer than a detailed standard, in my experience.
I never understood the first complaint. Yes, you can make = or + or whatever do crazy thing you like, but that goes for every function call. I remember debugging some (Java code, but that is besides the point) where the programmer had decided that getFoo() should actually change the internal state of the object in such a way that calling getFoo() twice totally screwed up the program. That was every bit as bad as if he has overloaded operator- to do the same.
Slicing, I agree, is a real problem, though I have seldom encountered it in real life. Perhaps because I like to keep all but my leaf classes abstract, which more or less eliminates the slicing problem.
This will be the, what, second standard? Third? And besides the new libraries, it's not that many new hard-to-grok features that are added. Actually, I can only think of lambda, but a language without lambda is an atrocity. Is there any particular feature you are thinking of? Have you read the wikipedia page?
I have worked on a number of projects, usually as some sort of project lead or technical lead. I have had problems with people who could not grok C++ (though not many), but I have not had the problem you describe, which just shows how lucky I am ;)
In my humble opinion the biggest problem I have had over the years is to get people to search for a freaking library instead of trying to do everything themselves. Admittedly, that is not a problem specific to C++.
Some of us wants them. For you that do not, pass std=c++98 to gcc and be happy. Think of it as a new language :)