In C++ programming, what librarys or frameworks are you using?

[CppNewer]

Or most peopel use the OS' API directly? or everyone made own series
of library such as Socket, Thread API.

 

[WuYaMing]

Or most peopel use the OS' API directly? or everyone made own series
of library such as Socket, Thread API.

gcc&g++

 

[Lars Uffmann]

Or most peopel use the OS' API directly? or everyone made own series
of library such as Socket, Thread API.

I'm currently using:

- wxWidgets (big difference to what I was used to, with the sizer
concept, but worth a look) with wxFormBuilder
- boost::thread
- winsock2 (supposed to be compatible to sockets-library under linux)
- xerces-c (domxml parsing)

all of this with Eclipse CDT & Cygwin, thinking about "migrating" to
MinGW though, for not needing cygwin1.dll or similar.

Best Regards,

Lars

 

[Erik Wikström]

Or most peopel use the OS' API directly? or everyone made own series
of library such as Socket, Thread API.

Depends on what kind of application I'm writing, I usually try to keep
the number of dependencies small. In the projects I'm currently working
on I use the C++ standard library, whatever I in POSIX, and Qt 3.

 

[Jeff Schwab]

Or most peopel use the OS' API directly? or everyone made own series
of library such as Socket, Thread API.

POSIX and Qt4.

You know what I would love? A great big honkin' printed tutorial on the
parts of Boost that could replace direct calls to POSIX. I would
particularly like portable access to the local file-system and to the
network, and portable concurrency. I'm aware that Boost provides these
libraries, but what I really want is an Addison-Wesley Professional
hardcover showing me how to use them. I guess that book won't exist
until TR2 makes it into C++1x.

 

[coal]

POSIX and Qt4.

You know what I would love?  A great big honkin' printed tutorial on the
parts of Boost that could replace direct calls to POSIX.  I would
particularly like portable access to the local file-system and to the
network, and portable concurrency.  I'm aware that Boost provides these
libraries, but what I really want is an Addison-Wesley Professional
hardcover showing me how to use them.  I guess that book won't exist
until TR2 makes it into C++1x.

I doubt it will take that long. I'm interested in hearing about
C++ related books that are in the works, but haven't been published
yet.

Brian Wood

 

[coal]

Or most peopel use the OS' API directly? or everyone made own series
of library such as Socket, Thread API.

Here's one related to sockets and marshalling -
www.webebenezer.net

Brian Wood

 

[EventHelix.com]

Or most peopel use the OS' API directly? or everyone made own series
of library such as Socket, Thread API.

ACE is C++ framework for network programming.

http://www.cs.wustl.edu/~schmidt/ACE.html

 

[davis685]

POSIX and Qt4.

You know what I would love? A great big honkin' printed tutorial on the
parts of Boost that could replace direct calls to POSIX. I would
particularly like portable access to the local file-system and to the
network, and portable concurrency. I'm aware that Boost provides these
libraries, but what I really want is an Addison-Wesley Professional
hardcover showing me how to use them. I guess that book won't exist
until TR2 makes it into C++1x.

I have been using this library (dclib.sourceforge.net) to do all of
that stuff portably and I think the documentation is pretty good. But
then again I wrote it

 

[Jeff Schwab]

I have been using this library (dclib.sourceforge.net) to do all of
that stuff portably and I think the documentation is pretty good. But
then again I wrote it

Thanks, that looks interesting. Now that std::tr1::array exists, are
you going to keep array_kernel_2?

 

[davis685]

Thanks, that looks interesting. Now that std::tr1::array exists, are
you going to keep array_kernel_2?

I probably will at least until C++0x is finalized and the tr1 stuff is
moved into the std namespace. The same goes for anything else that
is redundant with things in dlib (like the dlib::shared_ptr).

 

[coal]

I have been using this library (dclib.sourceforge.net) to do all of
that stuff portably and I think thedocumentationis pretty good.  But
then again I wrote it

On this page
http://dclib.sourceforge.net/dlib/serialize.h.html
it says, "Note that you should only try to deserialize an object to
the type of object it was serialized from. I.e. don't try anything
like loading a set with the serialized data of a queue." Why do you
impose that restriction? I don't have an application where I want
that functionality, but I've thought that it would probably be useful
at some point.

Brian Wood

 

[Gerry Ford]

Will you elaborate?

--
Gerry Ford

"Er hat sich georgiert." Der Spiegel, 2008, sich auf Chimpy Eins komma null
beziehend.

I have been using this library (dclib.sourceforge.net) to do all of
that stuff portably and I think thedocumentationis pretty good. But
then again I wrote it

On this page
http://dclib.sourceforge.net/dlib/serialize.h.html
it says, "Note that you should only try to deserialize an object to
the type of object it was serialized from. I.e. don't try anything
like loading a set with the serialized data of a queue." Why do you
impose that restriction? I don't have an application where I want
that functionality, but I've thought that it would probably be useful
at some point.

Brian Wood

 

[davis685]

Will you elaborate?

--
Gerry Ford

"Er hat sich georgiert." Der Spiegel, 2008, sich auf Chimpy Eins komma null



On this pagehttp://dclib.sourceforge.net/dlib/serialize.h.html
it says, "Note that you should only try to deserialize an object to
the type of object it was serialized from. I.e. don't try anything
like loading a set with the serialized data of a queue." Why do you
impose that restriction? I don't have an application where I want
that functionality, but I've thought that it would probably be useful
at some point.

Brian Wood

Well, the set object requires that every element in it be unique (i.e.
it isn't a multiset) so there are valid queue objects that can't be
loaded into sets. So all I'm saying is that if you are doing stuff
like that it isn't guaranteed to work since the source object might
violate some requirement imposed by the destination object type.
However, If you had a queue of unique integers then you could load
that into a set via deserialize without any problem.

Perhaps I should make that part of the documentation more clear.

-Davis

 

[coal]

Well, the set object requires that every element in it be unique (i.e.
it isn't a multiset) so there are valid queue objects that can't be
loaded into sets.

OK, but I wouldn't say it as "can't be loaded into sets." Data from a
queue<int> can be loaded into a set<int>, but the size of the set
might be less than the size of the queue. As long as users are aware
of that I think things should be fine. A documentation note as you
mention, could warn people to consider this possibility.

Brian Wood