The code quality was extremely bad, last time I had to work with it, must have been around 2012. Starting with hundreds of places where files did not get closed, then exceptions that did not get handled. I guess it is supposed to be used in a kind of batch process were it is restarted or only started to "import"/analyze files, so that crashes don't matter. We had it included as a library, until we ditched it.
Very true, I only read open source source code if there is a bug I need to maneauver around or fix. And most code is so bad, you don't really want to read it because of the night mares they induce, e.g. looking at https://lucene.apache.org/
Your argument is "the language doesnt need a for loop construct because it has if's and goto's. if's and goto's rule!" This is not my argument, as gotos and templates have nothing really to do with each other. If you don't have a while/repeat/for loop and you only have goto, you come not around using it. No one is forcing you to use templates if you really think they are not worth it.
I think C++ is losing ground for various other reasons: o Static compiled, long chain of included header files. o steep learning curve, probably coming form C makes it even harder o lack of a standard portable GUI library, well we have Qt, which is half commercial half OSS o much more typing for simple things than in Java o Smart pointers, that don't shield you from knowing what pointers actually are (and how they work) o weak support for unicode/strings different than char* o no reflection, which means no serialization o no VM, which means no load time or compile time byte code weaving, which means no aspect oriented programming or similar things (Yes, there is an aspect oriented extension/tool chain based on LLVM) o still no working cross linking of object files compiled by different compilers (on the same platform, don't even want to think about cross platform) o no remote function/method calls, unless you switch to CORBA and good bye to multiple inheritance then
Actually it is not really dead, as we still have Delphi/Embecado and http://freepascal.org/ (portable across many platforms)
I guess the reason of its decline is that e.g. Borland, which was big in Pascal, emphasized more on C++ and IDEs and bought a CASE company. Microsoft itself never was big in Pascal but focused on MFC. Apple switched from Pascal to C++ and then swiftly to Objective C.
For me "object oriented" Pascal always felt more like a hack than a sound object system. C++ is simply much more powerful, with operator overloading, multiple inheritance and templates.
In the late 1980s and early 1990s folks tried to promote Ada and Eiffel as new languages (they look a bit similar to Pascal on the first glance, Ada more than Eiffel) but both failed because they thought "programming languages" are "a product", and both started with incredible expensive compilers/IDEs.
And around 1995 Java came, and a few years later C# and.Net.
Pascal, like basically any other language, was once the most used programming language on the planet. E.g. everything around UCSD Pascal, later Turbo Pascal, then Apple/Mac Pascal, Delphi (yes, that is Pascal) and Embercardo, the actual Delphi Pascal.
There's nothing you can do in C++ that you can't do in C Then you don't know much about C++. Hint: constructors, destructors, custom allocators... you have noting like this in C. Oh, you mean you can write your own library in C and then manually call the functions? That is not what "languages" or "abstractions" are about.
Obviously you can write any program in any language, but then again: That is not what "languages" or "abstractions" are about.
there's dedicated language supporting this: D. It was made by C++ programmers who discovered this new template metaprogramming dimension in the first place. D is a child of Walter Bright, he is a C++ compiler writer, not a "discoverer" of TMP. TMP is partly discovered by the guys behind the STL, partly by Andrew Koenigs, James Coplien, Jiri Sirkoup, Ulrich Eisenecker and his PhD son, Krysztof Czarnecki (and partly me, but I only made one publication:D well, I guess there are plenty more people involved, but I never heard that Walter was deep into templates. But his Symantec C++ compiler/IDE - especially the graphical debugger - rocked big time)
C++ was originally intended to make the code easier to maintain, but like all efforts in that direction, the cost was in performance I doubt you find a convincing example where C++ code is less performant than C. Perhaps you can find a compiler or linker that compiles or links the former slower than the later, but that would be it./me facepalms
C is no longer the performance king. It never was. C is a small VM abstracting away registers and actually memory.
Modern programmers should be required to demonstrate proficiency with template meta programming as a prerequisite to being allowed to practice that is IMHO to much to ask for. I actually only know about 10, perhaps only 5, programmers who actually used those features.
much the way a course in lisp is included in virtually every accredited computer science degree. LISP is in my eyes a pain in the ass... people who like it most likely never will really do C++, but on the other hand might grasp template meta programming instantly: because it is the same as lisp, with different syntax/result.
If you treat it as C plus objects and stay away from operator overloading, stick with char arrays instead of strings, etc. If you do that, you are an idiot. Or you lack an ANSI C compiler, which you should use instead.
"Stay away fro operator overloading".../me shakes head
The destructor is a convenient place to update unwind semantics at all call sites
So is a macro definition, though. How does a Macro that you use at entry point of a function know that it need to do something when the function either returns or is rolled back by an exception?
Slashdot Mirror
RAM is extremely useful to cache the files you are working on ...
Swift and GCC both can compile on windows for Mac OS X/macOS.
https://www.makeuseof.com/tag/...
But why would you want to do that? Most likely it makes more sense to run the windows stuff you *need* in a VM on a Mac ...
I guess he wanted to be cynical.
absorbed enough radiation that is unhealthy.
A laser does not produce any radiation, except the laser beam.
For radiation you would need extremely odd lasers like a nuke induced gamma laser, or a free electron laser with stray X-Rays.
And to know that: you don't need to be a "laser safety officer".
I guess most Mac "Pro Users" are Java developers ;D
... I was hoping to find the password here, so I can fix my Abraham tank myself :(
The code quality was extremely bad, last time I had to work with it, must have been around 2012.
Starting with hundreds of places where files did not get closed, then exceptions that did not get handled.
I guess it is supposed to be used in a kind of batch process were it is restarted or only started to "import"/analyze files, so that crashes don't matter. We had it included as a library, until we ditched it.
Very true, I only read open source source code if there is a bug I need to maneauver around or fix.
And most code is so bad, you don't really want to read it because of the night mares they induce, e.g. looking at https://lucene.apache.org/
The connecting plug perhaps. The station not.
Electric power is electric power.
Your argument is "the language doesnt need a for loop construct because it has if's and goto's. if's and goto's rule!"
This is not my argument, as gotos and templates have nothing really to do with each other. If you don't have a while/repeat/for loop and you only have goto, you come not around using it. No one is forcing you to use templates if you really think they are not worth it.
I think C++ is losing ground for various other reasons:
o Static compiled, long chain of included header files.
o steep learning curve, probably coming form C makes it even harder
o lack of a standard portable GUI library, well we have Qt, which is half commercial half OSS
o much more typing for simple things than in Java
o Smart pointers, that don't shield you from knowing what pointers actually are (and how they work)
o weak support for unicode/strings different than char*
o no reflection, which means no serialization
o no VM, which means no load time or compile time byte code weaving, which means no aspect oriented programming or similar things (Yes, there is an aspect oriented extension/tool chain based on LLVM)
o still no working cross linking of object files compiled by different compilers (on the same platform, don't even want to think about cross platform)
o no remote function/method calls, unless you switch to CORBA and good bye to multiple inheritance then
A 5C temperature gradient you can have nearly everywhere.
A solar cell only works where you habe light, obviously.
Why don't you just download the data that you are interested in and check yourself?
Ha ha ha ha.
In what kind of dream worlddo you live?
A large block of memory is not a variable.
int a; <- that is a variable, and it is initialized to zero
To have uninitialized "variables" you need to define a struct/class that has no constructor.
An empty constructor initializes everything to 0 or 0.0.
Here is an interesting conversation about it: https://stackoverflow.com/ques...
Ah, I did not know that Andrei is now in the D team, I actually did not watch D that closely recent years.
Actually it is not really dead, as we still have Delphi/Embecado and http://freepascal.org/ (portable across many platforms)
I guess the reason of its decline is that e.g. Borland, which was big in Pascal, emphasized more on C++ and IDEs and bought a CASE company. Microsoft itself never was big in Pascal but focused on MFC. Apple switched from Pascal to C++ and then swiftly to Objective C.
For me "object oriented" Pascal always felt more like a hack than a sound object system. C++ is simply much more powerful, with operator overloading, multiple inheritance and templates.
In the late 1980s and early 1990s folks tried to promote Ada and Eiffel as new languages (they look a bit similar to Pascal on the first glance, Ada more than Eiffel) but both failed because they thought "programming languages" are "a product", and both started with incredible expensive compilers/IDEs.
And around 1995 Java came, and a few years later C# and .Net.
Pascal, like basically any other language, was once the most used programming language on the planet.
E.g. everything around UCSD Pascal, later Turbo Pascal, then Apple/Mac Pascal, Delphi (yes, that is Pascal) and Embercardo, the actual Delphi Pascal.
There's nothing you can do in C++ that you can't do in C ... you have noting like this in C.
Then you don't know much about C++.
Hint: constructors, destructors, custom allocators
Oh, you mean you can write your own library in C and then manually call the functions?
That is not what "languages" or "abstractions" are about.
Obviously you can write any program in any language, but then again: That is not what "languages" or "abstractions" are about.
there's dedicated language supporting this: D. It was made by C++ programmers who discovered this new template metaprogramming dimension in the first place. :D well, I guess there are plenty more people involved, but I never heard that Walter was deep into templates. But his Symantec C++ compiler/IDE - especially the graphical debugger - rocked big time)
D is a child of Walter Bright, he is a C++ compiler writer, not a "discoverer" of TMP. TMP is partly discovered by the guys behind the STL, partly by Andrew Koenigs, James Coplien, Jiri Sirkoup, Ulrich Eisenecker and his PhD son, Krysztof Czarnecki (and partly me, but I only made one publication
C++ was originally intended to make the code easier to maintain, but like all efforts in that direction, the cost was in performance /me facepalms
I doubt you find a convincing example where C++ code is less performant than C.
Perhaps you can find a compiler or linker that compiles or links the former slower than the later, but that would be it.
C is no longer the performance king.
It never was. C is a small VM abstracting away registers and actually memory.
Modern programmers should be required to demonstrate proficiency with template meta programming as a prerequisite to being allowed to practice that is IMHO to much to ask for. I actually only know about 10, perhaps only 5, programmers who actually used those features.
much the way a course in lisp is included in virtually every accredited computer science degree. ... people who like it most likely never will really do C++, but on the other hand might grasp template meta programming instantly: because it is the same as lisp, with different syntax/result.
LISP is in my eyes a pain in the ass
uninitialized variables/memory, etc. ...
C++ does not have uninitialized variables
Coding anything mathematical in Java was an exercise in frustration as a direct result of making Java "safe".
That is why we use Groovy and/or Scala.
If you treat it as C plus objects and stay away from operator overloading, stick with char arrays instead of strings, etc.
If you do that, you are an idiot. Or you lack an ANSI C compiler, which you should use instead.
"Stay away fro operator overloading" ... /me shakes head
The biggest mistakes of Java and C# are that they favoured generics over templates.
No idea why other people think otherwise. Template specialization, template meta programming: all not possible with generics.
The destructor is a convenient place to update unwind semantics at all call sites
So is a macro definition, though.
How does a Macro that you use at entry point of a function know that it need to do something when the function either returns or is rolled back by an exception?