This is a list of changes to libwww in reverse chronological order. See also the release notes for the latest release.
This is a minor release that mainly provides bug fixes and small enhancements that have been commited to CVS since the last release. Vladimir Koslov provided indispensable help for testing and preparing the Win32 release. Special thanks to Wayne Davison, Art Barstow, Peter Stamfest, Zhu Qun-Ying, Jens Meggers, Ken Olum for their contributions.
head.c, getheaders.c, chunk.c, chunkbody.c,
multichunk.c
When building libwww on Windows 95/98/NT, you can decide on a set of different options to be enabled or disabled:
Non-blocking network access can either be done using multiplexed I/O or
asynchronous I/O. Multiplexed I/O is based on an event loop with a
select() system call. The select() system call is a
function that scans a set of BSD like socket descriptors and returns when one
or more sockets are ready for either network read or write.
Asynchronous I/O is based on signals instead of a select() call.
Each time a socket is ready for a network operation, for example read,
a signal is generated. This signal can then be caught by the application in an
signal handler. Using the select() call is called
reactive whereas using signals is called pro-active.
Libwww can handle both approaches under Windows, and therefore you must choose which model you prefer. The choice can depend on what type of application you are using and what other libraries you are using.
In this mode the Event Manager registers
all active sockets and passes them to a select() call which then
processes the registered sockets. When select() returns the Event
Manager dispatches the appropriate handlers as the sockets get ready.
What to do:
Undefine WWW_WIN_ASYNC as a preprocessor directive when
generating the project for the Library. Please make sure that all
DLLs are compiled without this flag.
Here the Event Manager registers all
active sockets using WSAAsyncSelect which is part of the WinSock
API. When AsyncSelect() which is the asynchronous equivalent to
the select() returns, the Event Manager dispatches the results of
the AsyncSelect(). As the asynchronous select call needs a
windows handle, the Library creates a hidden window. This window is
not to be used explicitly by the application and the Library closes
the window when exiting.
What to do:
Define WWW_WIN_ASYNC as preprocessor directives when
generating the project for the Library.
A Windows application can use either a character based command line interface, or a graphic windows interface. Many MS-DOS applications do run as a simple command line tool which doesn't use GUI at all. Under Win32, there is a notion of a console application. This means that all user interaction happens through a standard DOS shell interface, with a FILE pointer like that in Unix. This does not exist under Win16, where a Windows window is required.
Libwww supports both the console and the windows interface, and again you must choose what version you prefer. This is often something you have to decide as you are creating the project.
If you want to make a windows application then this is the mode to use. Eric Prud'hommeaux has provided a Windows application wrapper for the W3C Line Mode Browser which you can use in order to build the browser as a Windows application. This is found in www.c. Three other modules, scroll.c and lib.c provide the window for the application.
What to do: Define
_WINDOWS as preprocessor directives when generating the project for the
Library.
The console option is available only in Win32 in which case all user interaction happens through the Win32 console window. This model strongly resembles a Unix vt100 terminal interface.
What to do: Define
_CONSOLE as preprocessor directives when generating the project for the
Library.
Windows has a concept of both static and dynamic libraries, the latter also known as DLLs. It is out of scope here to describe the difference between DLLs and static libraries, but as DLLs is based on a lot more flexible memory model it is almost always the best solution for Windows applications.
The W3C Sample Code Library support both models in that if can be built as either one big static library or as a set of small DLLs. As mentioned, it is in almost all cases recommended to build DLLs instead of static libraries, and on Win16 it is required because a static library is too big.
The libraries may be build as one large static library. This is how libwww is
implemented on Unix platforms. Subsequent references to the various DLLs may
all be assumed to refer to the staticly linked libwww. Care has been taken to
insure that there are no #define conflicts where one library
would want a #define that would interfere with the modules in
other libraries. When building a static library, see the following sections on
Select Options and Input/Output
Options, accumulate all the #defines that are required, and
build the whole of libwww with those #defines.
What to do:
Undefine WWW_WIN_DLL as preprocessor directives when
generating the project for the Library. Please note, that it is not
recommended to staticly link to the libraries if you are building a Win16
application as it creates segment size problems.
The libwww can also be built as a set of DLLs that follows the modular architecture of the Library. This enables the application programmer to choose exactly what functionality should be enabled in the application. The boundaries between these DLLs are based on module interdependency and some assumptions regarding which modules may be replaced by the application. Unlike static linking, dynamic linking requires that all the modules in a DLL be replaced at once. This is because the DLL needs all internal references to be resolved at build time.
What to do: Define
WWW_WIN_DLL as preprocessor directives when generating the project for
the Library.
In addition to platform independent modules, there is a small Windows specific
DLL which implements the trace message generation. The DLL is called
windll.dll and contains also the definition of the global trace
flag definition.
windll.c - DLLMain to all DLLs.
wwwdll.c - instances WWW_TraceFlag. This is included
in wwwdll so that it can export the variables to the rest of the DLLs.
wwwdll.def, wwwutils.def, wwwcore.def -
def (exports) files for the DLLs of the same names.
What to do: Include
the windll.dll as a part of your project and make sure that it is built as the
first DLL.
The functions exported from a DLL are listed in the EXPORTS
section of a .def file. These can be found in the windows directory. These may also be build by the
makeDefs.pl perl script, see the
description below. You can use the def files as a basis for generating the DLL
projects for your compiler.
What to do: Generate
the DLLs according to the def files so the exported interface is identical to
the set of functions defined in the actual c files included in the DLL.
Unfortunately, make files are not easily shared among different C compilers on Windows which complicates the distribution. See the installation instructions for details on possible ways of building libwww on windows.
This list gives an overview of the platforms where libwww is known (or belived) to work. Thanks a lot to the big group of creative contributors for the long list of hints, patches and additions to the make files and system dependent code! If you have any problems or questions then send a mail to the www-lib@w3.org mailing list (read the instructions for how to subscribe). We also keep an archive of the mailing list so that more easily can get up to speed.
| Platform | OS | Note |
|---|---|---|
| Unix | Tried on Solaris, Linux and Digital Unix | Build the library using GNU Autoconf's configure script as described in the installation instructions. Note: You must use GNU make in order to do this |
| Windows | Tried on Windows 95/98/NT | Follow the installation instructions and see the special notes for building on Windows |
| Macs | MacOS | It doesn't work out of the box but several people are using it on Macs. Ask on the <www-lib@w3.org> mailing list |
Disclaimer: Libwww has been designed to be both portable and flexible based on ANSI C and POSIX. However, there are platforms that do not conform to these standards. Even though autoconf is often capable of getting around such problems it may not always be the case. In this situation we are happy to accept patches but do not promise to maintain any platforms other than the ones we use for development.
This is a list of changes to libwww in reverse chronological order. See also the release notes for the latest release.
This is a minor release that mainly provides bug fixes and small enhancements that have been commited to CVS since the last release. Vladimir Koslov provided indispensable help for testing and preparing the Win32 release. Special thanks to Wayne Davison, Art Barstow, Peter Stamfest, Zhu Qun-Ying, Jens Meggers, Ken Olum for their contributions.
head.c, getheaders.c, chunk.c, chunkbody.c,
multichunk.c
When building libwww on Windows 95/98/NT, you can decide on a set of different options to be enabled or disabled:
Non-blocking network access can either be done using multiplexed I/O or
asynchronous I/O. Multiplexed I/O is based on an event loop with a
select() system call. The select() system call is a
function that scans a set of BSD like socket descriptors and returns when one
or more sockets are ready for either network read or write.
Asynchronous I/O is based on signals instead of a select() call.
Each time a socket is ready for a network operation, for example read,
a signal is generated. This signal can then be caught by the application in an
signal handler. Using the select() call is called
reactive whereas using signals is called pro-active.
Libwww can handle both approaches under Windows, and therefore you must choose which model you prefer. The choice can depend on what type of application you are using and what other libraries you are using.
In this mode the Event Manager registers
all active sockets and passes them to a select() call which then
processes the registered sockets. When select() returns the Event
Manager dispatches the appropriate handlers as the sockets get ready.
What to do:
Undefine WWW_WIN_ASYNC as a preprocessor directive when
generating the project for the Library. Please make sure that all
DLLs are compiled without this flag.
Here the Event Manager registers all
active sockets using WSAAsyncSelect which is part of the WinSock
API. When AsyncSelect() which is the asynchronous equivalent to
the select() returns, the Event Manager dispatches the results of
the AsyncSelect(). As the asynchronous select call needs a
windows handle, the Library creates a hidden window. This window is
not to be used explicitly by the application and the Library closes
the window when exiting.
What to do:
Define WWW_WIN_ASYNC as preprocessor directives when
generating the project for the Library.
A Windows application can use either a character based command line interface, or a graphic windows interface. Many MS-DOS applications do run as a simple command line tool which doesn't use GUI at all. Under Win32, there is a notion of a console application. This means that all user interaction happens through a standard DOS shell interface, with a FILE pointer like that in Unix. This does not exist under Win16, where a Windows window is required.
Libwww supports both the console and the windows interface, and again you must choose what version you prefer. This is often something you have to decide as you are creating the project.
If you want to make a windows application then this is the mode to use. Eric Prud'hommeaux has provided a Windows application wrapper for the W3C Line Mode Browser which you can use in order to build the browser as a Windows application. This is found in www.c. Three other modules, scroll.c and lib.c provide the window for the application.
What to do: Define
_WINDOWS as preprocessor directives when generating the project for the
Library.
The console option is available only in Win32 in which case all user interaction happens through the Win32 console window. This model strongly resembles a Unix vt100 terminal interface.
What to do: Define
_CONSOLE as preprocessor directives when generating the project for the
Library.
Windows has a concept of both static and dynamic libraries, the latter also known as DLLs. It is out of scope here to describe the difference between DLLs and static libraries, but as DLLs is based on a lot more flexible memory model it is almost always the best solution for Windows applications.
The W3C Sample Code Library support both models in that if can be built as either one big static library or as a set of small DLLs. As mentioned, it is in almost all cases recommended to build DLLs instead of static libraries, and on Win16 it is required because a static library is too big.
The libraries may be build as one large static library. This is how libwww is
implemented on Unix platforms. Subsequent references to the various DLLs may
all be assumed to refer to the staticly linked libwww. Care has been taken to
insure that there are no #define conflicts where one library
would want a #define that would interfere with the modules in
other libraries. When building a static library, see the following sections on
Select Options and Input/Output
Options, accumulate all the #defines that are required, and
build the whole of libwww with those #defines.
What to do:
Undefine WWW_WIN_DLL as preprocessor directives when
generating the project for the Library. Please note, that it is not
recommended to staticly link to the libraries if you are building a Win16
application as it creates segment size problems.
The libwww can also be built as a set of DLLs that follows the modular architecture of the Library. This enables the application programmer to choose exactly what functionality should be enabled in the application. The boundaries between these DLLs are based on module interdependency and some assumptions regarding which modules may be replaced by the application. Unlike static linking, dynamic linking requires that all the modules in a DLL be replaced at once. This is because the DLL needs all internal references to be resolved at build time.
What to do: Define
WWW_WIN_DLL as preprocessor directives when generating the project for
the Library.
In addition to platform independent modules, there is a small Windows specific
DLL which implements the trace message generation. The DLL is called
windll.dll and contains also the definition of the global trace
flag definition.
windll.c - DLLMain to all DLLs.
wwwdll.c - instances WWW_TraceFlag. This is included
in wwwdll so that it can export the variables to the rest of the DLLs.
wwwdll.def, wwwutils.def, wwwcore.def -
def (exports) files for the DLLs of the same names.
What to do: Include
the windll.dll as a part of your project and make sure that it is built as the
first DLL.
The functions exported from a DLL are listed in the EXPORTS
section of a .def file. These can be found in the windows directory. These may also be build by the
makeDefs.pl perl script, see the
description below. You can use the def files as a basis for generating the DLL
projects for your compiler.
What to do: Generate
the DLLs according to the def files so the exported interface is identical to
the set of functions defined in the actual c files included in the DLL.
Unfortunately, make files are not easily shared among different C compilers on Windows which complicates the distribution. See the installation instructions for details on possible ways of building libwww on windows.
This list gives an overview of the platforms where libwww is known (or belived) to work. Thanks a lot to the big group of creative contributors for the long list of hints, patches and additions to the make files and system dependent code! If you have any problems or questions then send a mail to the www-lib@w3.org mailing list (read the instructions for how to subscribe). We also keep an archive of the mailing list so that more easily can get up to speed.
| Platform | OS | Note |
|---|---|---|
| Unix | Tried on Solaris, Linux and Digital Unix | Build the library using GNU Autoconf's configure script as described in the installation instructions. Note: You must use GNU make in order to do this |
| Windows | Tried on Windows 95/98/NT | Follow the installation instructions and see the special notes for building on Windows |
| Macs | MacOS | It doesn't work out of the box but several people are using it on Macs. Ask on the <www-lib@w3.org> mailing list |
Disclaimer: Libwww has been designed to be both portable and flexible based on ANSI C and POSIX. However, there are platforms that do not conform to these standards. Even though autoconf is often capable of getting around such problems it may not always be the case. In this situation we are happy to accept patches but do not promise to maintain any platforms other than the ones we use for development.
This is a list of changes to libwww in reverse chronological order. See also the release notes for the latest release.
This is a minor release that mainly provides bug fixes and small enhancements that have been commited to CVS since the last release. Vladimir Koslov provided indispensable help for testing and preparing the Win32 release. Special thanks to Wayne Davison, Art Barstow, Peter Stamfest, Zhu Qun-Ying, Jens Meggers, Ken Olum for their contributions.
head.c, getheaders.c, chunk.c, chunkbody.c,
multichunk.c
When building libwww on Windows 95/98/NT, you can decide on a set of different options to be enabled or disabled:
Non-blocking network access can either be done using multiplexed I/O or
asynchronous I/O. Multiplexed I/O is based on an event loop with a
select() system call. The select() system call is a
function that scans a set of BSD like socket descriptors and returns when one
or more sockets are ready for either network read or write.
Asynchronous I/O is based on signals instead of a select() call.
Each time a socket is ready for a network operation, for example read,
a signal is generated. This signal can then be caught by the application in an
signal handler. Using the select() call is called
reactive whereas using signals is called pro-active.
Libwww can handle both approaches under Windows, and therefore you must choose which model you prefer. The choice can depend on what type of application you are using and what other libraries you are using.
In this mode the Event Manager registers
all active sockets and passes them to a select() call which then
processes the registered sockets. When select() returns the Event
Manager dispatches the appropriate handlers as the sockets get ready.
What to do:
Undefine WWW_WIN_ASYNC as a preprocessor directive when
generating the project for the Library. Please make sure that all
DLLs are compiled without this flag.
Here the Event Manager registers all
active sockets using WSAAsyncSelect which is part of the WinSock
API. When AsyncSelect() which is the asynchronous equivalent to
the select() returns, the Event Manager dispatches the results of
the AsyncSelect(). As the asynchronous select call needs a
windows handle, the Library creates a hidden window. This window is
not to be used explicitly by the application and the Library closes
the window when exiting.
What to do:
Define WWW_WIN_ASYNC as preprocessor directives when
generating the project for the Library.
A Windows application can use either a character based command line interface, or a graphic windows interface. Many MS-DOS applications do run as a simple command line tool which doesn't use GUI at all. Under Win32, there is a notion of a console application. This means that all user interaction happens through a standard DOS shell interface, with a FILE pointer like that in Unix. This does not exist under Win16, where a Windows window is required.
Libwww supports both the console and the windows interface, and again you must choose what version you prefer. This is often something you have to decide as you are creating the project.
If you want to make a windows application then this is the mode to use. Eric Prud'hommeaux has provided a Windows application wrapper for the W3C Line Mode Browser which you can use in order to build the browser as a Windows application. This is found in www.c. Three other modules, scroll.c and lib.c provide the window for the application.
What to do: Define
_WINDOWS as preprocessor directives when generating the project for the
Library.
The console option is available only in Win32 in which case all user interaction happens through the Win32 console window. This model strongly resembles a Unix vt100 terminal interface.
What to do: Define
_CONSOLE as preprocessor directives when generating the project for the
Library.
Windows has a concept of both static and dynamic libraries, the latter also known as DLLs. It is out of scope here to describe the difference between DLLs and static libraries, but as DLLs is based on a lot more flexible memory model it is almost always the best solution for Windows applications.
The W3C Sample Code Library support both models in that if can be built as either one big static library or as a set of small DLLs. As mentioned, it is in almost all cases recommended to build DLLs instead of static libraries, and on Win16 it is required because a static library is too big.
The libraries may be build as one large static library. This is how libwww is
implemented on Unix platforms. Subsequent references to the various DLLs may
all be assumed to refer to the staticly linked libwww. Care has been taken to
insure that there are no #define conflicts where one library
would want a #define that would interfere with the modules in
other libraries. When building a static library, see the following sections on
Select Options and Input/Output
Options, accumulate all the #defines that are required, and
build the whole of libwww with those #defines.
What to do:
Undefine WWW_WIN_DLL as preprocessor directives when
generating the project for the Library. Please note, that it is not
recommended to staticly link to the libraries if you are building a Win16
application as it creates segment size problems.
The libwww can also be built as a set of DLLs that follows the modular architecture of the Library. This enables the application programmer to choose exactly what functionality should be enabled in the application. The boundaries between these DLLs are based on module interdependency and some assumptions regarding which modules may be replaced by the application. Unlike static linking, dynamic linking requires that all the modules in a DLL be replaced at once. This is because the DLL needs all internal references to be resolved at build time.
What to do: Define
WWW_WIN_DLL as preprocessor directives when generating the project for
the Library.
In addition to platform independent modules, there is a small Windows specific
DLL which implements the trace message generation. The DLL is called
windll.dll and contains also the definition of the global trace
flag definition.
windll.c - DLLMain to all DLLs.
wwwdll.c - instances WWW_TraceFlag. This is included
in wwwdll so that it can export the variables to the rest of the DLLs.
wwwdll.def, wwwutils.def, wwwcore.def -
def (exports) files for the DLLs of the same names.
What to do: Include
the windll.dll as a part of your project and make sure that it is built as the
first DLL.
The functions exported from a DLL are listed in the EXPORTS
section of a .def file. These can be found in the windows directory. These may also be build by the
makeDefs.pl perl script, see the
description below. You can use the def files as a basis for generating the DLL
projects for your compiler.
What to do: Generate
the DLLs according to the def files so the exported interface is identical to
the set of functions defined in the actual c files included in the DLL.
Unfortunately, make files are not easily shared among different C compilers on Windows which complicates the distribution. See the installation instructions for details on possible ways of building libwww on windows.
This list gives an overview of the platforms where libwww is known (or belived) to work. Thanks a lot to the big group of creative contributors for the long list of hints, patches and additions to the make files and system dependent code! If you have any problems or questions then send a mail to the www-lib@w3.org mailing list (read the instructions for how to subscribe). We also keep an archive of the mailing list so that more easily can get up to speed.
| Platform | OS | Note |
|---|---|---|
| Unix | Tried on Solaris, Linux and Digital Unix | Build the library using GNU Autoconf's configure script as described in the installation instructions. Note: You must use GNU make in order to do this |
| Windows | Tried on Windows 95/98/NT | Follow the installation instructions and see the special notes for building on Windows |
| Macs | MacOS | It doesn't work out of the box but several people are using it on Macs. Ask on the <www-lib@w3.org> mailing list |
Disclaimer: Libwww has been designed to be both portable and flexible based on ANSI C and POSIX. However, there are platforms that do not conform to these standards. Even though autoconf is often capable of getting around such problems it may not always be the case. In this situation we are happy to accept patches but do not promise to maintain any platforms other than the ones we use for development.
This is a list of changes to libwww in reverse chronological order. See also the release notes for the latest release.
This is a minor release that mainly provides bug fixes and small enhancements that have been commited to CVS since the last release. Vladimir Koslov provided indispensable help for testing and preparing the Win32 release. Special thanks to Wayne Davison, Art Barstow, Peter Stamfest, Zhu Qun-Ying, Jens Meggers, Ken Olum for their contributions.
head.c, getheaders.c, chunk.c, chunkbody.c,
multichunk.c
When building libwww on Windows 95/98/NT, you can decide on a set of different options to be enabled or disabled:
Non-blocking network access can either be done using multiplexed I/O or
asynchronous I/O. Multiplexed I/O is based on an event loop with a
select() system call. The select() system call is a
function that scans a set of BSD like socket descriptors and returns when one
or more sockets are ready for either network read or write.
Asynchronous I/O is based on signals instead of a select() call.
Each time a socket is ready for a network operation, for example read,
a signal is generated. This signal can then be caught by the application in an
signal handler. Using the select() call is called
reactive whereas using signals is called pro-active.
Libwww can handle both approaches under Windows, and therefore you must choose which model you prefer. The choice can depend on what type of application you are using and what other libraries you are using.
In this mode the Event Manager registers
all active sockets and passes them to a select() call which then
processes the registered sockets. When select() returns the Event
Manager dispatches the appropriate handlers as the sockets get ready.
What to do:
Undefine WWW_WIN_ASYNC as a preprocessor directive when
generating the project for the Library. Please make sure that all
DLLs are compiled without this flag.
Here the Event Manager registers all
active sockets using WSAAsyncSelect which is part of the WinSock
API. When AsyncSelect() which is the asynchronous equivalent to
the select() returns, the Event Manager dispatches the results of
the AsyncSelect(). As the asynchronous select call needs a
windows handle, the Library creates a hidden window. This window is
not to be used explicitly by the application and the Library closes
the window when exiting.
What to do:
Define WWW_WIN_ASYNC as preprocessor directives when
generating the project for the Library.
A Windows application can use either a character based command line interface, or a graphic windows interface. Many MS-DOS applications do run as a simple command line tool which doesn't use GUI at all. Under Win32, there is a notion of a console application. This means that all user interaction happens through a standard DOS shell interface, with a FILE pointer like that in Unix. This does not exist under Win16, where a Windows window is required.
Libwww supports both the console and the windows interface, and again you must choose what version you prefer. This is often something you have to decide as you are creating the project.
If you want to make a windows application then this is the mode to use. Eric Prud'hommeaux has provided a Windows application wrapper for the W3C Line Mode Browser which you can use in order to build the browser as a Windows application. This is found in www.c. Three other modules, scroll.c and lib.c provide the window for the application.
What to do: Define
_WINDOWS as preprocessor directives when generating the project for the
Library.
The console option is available only in Win32 in which case all user interaction happens through the Win32 console window. This model strongly resembles a Unix vt100 terminal interface.
What to do: Define
_CONSOLE as preprocessor directives when generating the project for the
Library.
Windows has a concept of both static and dynamic libraries, the latter also known as DLLs. It is out of scope here to describe the difference between DLLs and static libraries, but as DLLs is based on a lot more flexible memory model it is almost always the best solution for Windows applications.
The W3C Sample Code Library support both models in that if can be built as either one big static library or as a set of small DLLs. As mentioned, it is in almost all cases recommended to build DLLs instead of static libraries, and on Win16 it is required because a static library is too big.
The libraries may be build as one large static library. This is how libwww is
implemented on Unix platforms. Subsequent references to the various DLLs may
all be assumed to refer to the staticly linked libwww. Care has been taken to
insure that there are no #define conflicts where one library
would want a #define that would interfere with the modules in
other libraries. When building a static library, see the following sections on
Select Options and Input/Output
Options, accumulate all the #defines that are required, and
build the whole of libwww with those #defines.
What to do:
Undefine WWW_WIN_DLL as preprocessor directives when
generating the project for the Library. Please note, that it is not
recommended to staticly link to the libraries if you are building a Win16
application as it creates segment size problems.
The libwww can also be built as a set of DLLs that follows the modular architecture of the Library. This enables the application programmer to choose exactly what functionality should be enabled in the application. The boundaries between these DLLs are based on module interdependency and some assumptions regarding which modules may be replaced by the application. Unlike static linking, dynamic linking requires that all the modules in a DLL be replaced at once. This is because the DLL needs all internal references to be resolved at build time.
What to do: Define
WWW_WIN_DLL as preprocessor directives when generating the project for
the Library.
In addition to platform independent modules, there is a small Windows specific
DLL which implements the trace message generation. The DLL is called
windll.dll and contains also the definition of the global trace
flag definition.
windll.c - DLLMain to all DLLs.
wwwdll.c - instances WWW_TraceFlag. This is included
in wwwdll so that it can export the variables to the rest of the DLLs.
wwwdll.def, wwwutils.def, wwwcore.def -
def (exports) files for the DLLs of the same names.
What to do: Include
the windll.dll as a part of your project and make sure that it is built as the
first DLL.
The functions exported from a DLL are listed in the EXPORTS
section of a .def file. These can be found in the windows directory. These may also be build by the
makeDefs.pl perl script, see the
description below. You can use the def files as a basis for generating the DLL
projects for your compiler.
What to do: Generate
the DLLs according to the def files so the exported interface is identical to
the set of functions defined in the actual c files included in the DLL.
Unfortunately, make files are not easily shared among different C compilers on Windows which complicates the distribution. See the installation instructions for details on possible ways of building libwww on windows.
This list gives an overview of the platforms where libwww is known (or belived) to work. Thanks a lot to the big group of creative contributors for the long list of hints, patches and additions to the make files and system dependent code! If you have any problems or questions then send a mail to the www-lib@w3.org mailing list (read the instructions for how to subscribe). We also keep an archive of the mailing list so that more easily can get up to speed.
| Platform | OS | Note |
|---|---|---|
| Unix | Tried on Solaris, Linux and Digital Unix | Build the library using GNU Autoconf's configure script as described in the installation instructions. Note: You must use GNU make in order to do this |
| Windows | Tried on Windows 95/98/NT | Follow the installation instructions and see the special notes for building on Windows |
| Macs | MacOS | It doesn't work out of the box but several people are using it on Macs. Ask on the <www-lib@w3.org> mailing list |
Disclaimer: Libwww has been designed to be both portable and flexible based on ANSI C and POSIX. However, there are platforms that do not conform to these standards. Even though autoconf is often capable of getting around such problems it may not always be the case. In this situation we are happy to accept patches but do not promise to maintain any platforms other than the ones we use for development.
This is a list of changes to libwww in reverse chronological order. See also the release notes for the latest release.
This is a minor release that mainly provides bug fixes and small enhancements that have been commited to CVS since the last release. Vladimir Koslov provided indispensable help for testing and preparing the Win32 release. Special thanks to Wayne Davison, Art Barstow, Peter Stamfest, Zhu Qun-Ying, Jens Meggers, Ken Olum for their contributions.
head.c, getheaders.c, chunk.c, chunkbody.c,
multichunk.c
When building libwww on Windows 95/98/NT, you can decide on a set of different options to be enabled or disabled:
Non-blocking network access can either be done using multiplexed I/O or
asynchronous I/O. Multiplexed I/O is based on an event loop with a
select() system call. The select() system call is a
function that scans a set of BSD like socket descriptors and returns when one
or more sockets are ready for either network read or write.
Asynchronous I/O is based on signals instead of a select() call.
Each time a socket is ready for a network operation, for example read,
a signal is generated. This signal can then be caught by the application in an
signal handler. Using the select() call is called
reactive whereas using signals is called pro-active.
Libwww can handle both approaches under Windows, and therefore you must choose which model you prefer. The choice can depend on what type of application you are using and what other libraries you are using.
In this mode the Event Manager registers
all active sockets and passes them to a select() call which then
processes the registered sockets. When select() returns the Event
Manager dispatches the appropriate handlers as the sockets get ready.
What to do:
Undefine WWW_WIN_ASYNC as a preprocessor directive when
generating the project for the Library. Please make sure that all
DLLs are compiled without this flag.
Here the Event Manager registers all
active sockets using WSAAsyncSelect which is part of the WinSock
API. When AsyncSelect() which is the asynchronous equivalent to
the select() returns, the Event Manager dispatches the results of
the AsyncSelect(). As the asynchronous select call needs a
windows handle, the Library creates a hidden window. This window is
not to be used explicitly by the application and the Library closes
the window when exiting.
What to do:
Define WWW_WIN_ASYNC as preprocessor directives when
generating the project for the Library.
A Windows application can use either a character based command line interface, or a graphic windows interface. Many MS-DOS applications do run as a simple command line tool which doesn't use GUI at all. Under Win32, there is a notion of a console application. This means that all user interaction happens through a standard DOS shell interface, with a FILE pointer like that in Unix. This does not exist under Win16, where a Windows window is required.
Libwww supports both the console and the windows interface, and again you must choose what version you prefer. This is often something you have to decide as you are creating the project.
If you want to make a windows application then this is the mode to use. Eric Prud'hommeaux has provided a Windows application wrapper for the W3C Line Mode Browser which you can use in order to build the browser as a Windows application. This is found in www.c. Three other modules, scroll.c and lib.c provide the window for the application.
What to do: Define
_WINDOWS as preprocessor directives when generating the project for the
Library.
The console option is available only in Win32 in which case all user interaction happens through the Win32 console window. This model strongly resembles a Unix vt100 terminal interface.
What to do: Define
_CONSOLE as preprocessor directives when generating the project for the
Library.
Windows has a concept of both static and dynamic libraries, the latter also known as DLLs. It is out of scope here to describe the difference between DLLs and static libraries, but as DLLs is based on a lot more flexible memory model it is almost always the best solution for Windows applications.
The W3C Sample Code Library support both models in that if can be built as either one big static library or as a set of small DLLs. As mentioned, it is in almost all cases recommended to build DLLs instead of static libraries, and on Win16 it is required because a static library is too big.
The libraries may be build as one large static library. This is how libwww is
implemented on Unix platforms. Subsequent references to the various DLLs may
all be assumed to refer to the staticly linked libwww. Care has been taken to
insure that there are no #define conflicts where one library
would want a #define that would interfere with the modules in
other libraries. When building a static library, see the following sections on
Select Options and Input/Output
Options, accumulate all the #defines that are required, and
build the whole of libwww with those #defines.
What to do:
Undefine WWW_WIN_DLL as preprocessor directives when
generating the project for the Library. Please note, that it is not
recommended to staticly link to the libraries if you are building a Win16
application as it creates segment size problems.
The libwww can also be built as a set of DLLs that follows the modular architecture of the Library. This enables the application programmer to choose exactly what functionality should be enabled in the application. The boundaries between these DLLs are based on module interdependency and some assumptions regarding which modules may be replaced by the application. Unlike static linking, dynamic linking requires that all the modules in a DLL be replaced at once. This is because the DLL needs all internal references to be resolved at build time.
What to do: Define
WWW_WIN_DLL as preprocessor directives when generating the project for
the Library.
In addition to platform independent modules, there is a small Windows specific
DLL which implements the trace message generation. The DLL is called
windll.dll and contains also the definition of the global trace
flag definition.
windll.c - DLLMain to all DLLs.
wwwdll.c - instances WWW_TraceFlag. This is included
in wwwdll so that it can export the variables to the rest of the DLLs.
wwwdll.def, wwwutils.def, wwwcore.def -
def (exports) files for the DLLs of the same names.
What to do: Include
the windll.dll as a part of your project and make sure that it is built as the
first DLL.
The functions exported from a DLL are listed in the EXPORTS
section of a .def file. These can be found in the windows directory. These may also be build by the
makeDefs.pl perl script, see the
description below. You can use the def files as a basis for generating the DLL
projects for your compiler.
What to do: Generate
the DLLs according to the def files so the exported interface is identical to
the set of functions defined in the actual c files included in the DLL.
Unfortunately, make files are not easily shared among different C compilers on Windows which complicates the distribution. See the installation instructions for details on possible ways of building libwww on windows.
This list gives an overview of the platforms where libwww is known (or belived) to work. Thanks a lot to the big group of creative contributors for the long list of hints, patches and additions to the make files and system dependent code! If you have any problems or questions then send a mail to the www-lib@w3.org mailing list (read the instructions for how to subscribe). We also keep an archive of the mailing list so that more easily can get up to speed.
| Platform | OS | Note |
|---|---|---|
| Unix | Tried on Solaris, Linux and Digital Unix | Build the library using GNU Autoconf's configure script as described in the installation instructions. Note: You must use GNU make in order to do this |
| Windows | Tried on Windows 95/98/NT | Follow the installation instructions and see the special notes for building on Windows |
| Macs | MacOS | It doesn't work out of the box but several people are using it on Macs. Ask on the <www-lib@w3.org> mailing list |
Disclaimer: Libwww has been designed to be both portable and flexible based on ANSI C and POSIX. However, there are platforms that do not conform to these standards. Even though autoconf is often capable of getting around such problems it may not always be the case. In this situation we are happy to accept patches but do not promise to maintain any platforms other than the ones we use for development.
This is a list of changes to libwww in reverse chronological order. See also the release notes for the latest release.
This is a minor release that mainly provides bug fixes and small enhancements that have been commited to CVS since the last release. Vladimir Koslov provided indispensable help for testing and preparing the Win32 release. Special thanks to Wayne Davison, Art Barstow, Peter Stamfest, Zhu Qun-Ying, Jens Meggers, Ken Olum for their contributions.
head.c, getheaders.c, chunk.c, chunkbody.c,
multichunk.c
When building libwww on Windows 95/98/NT, you can decide on a set of different options to be enabled or disabled:
Non-blocking network access can either be done using multiplexed I/O or
asynchronous I/O. Multiplexed I/O is based on an event loop with a
select() system call. The select() system call is a
function that scans a set of BSD like socket descriptors and returns when one
or more sockets are ready for either network read or write.
Asynchronous I/O is based on signals instead of a select() call.
Each time a socket is ready for a network operation, for example read,
a signal is generated. This signal can then be caught by the application in an
signal handler. Using the select() call is called
reactive whereas using signals is called pro-active.
Libwww can handle both approaches under Windows, and therefore you must choose which model you prefer. The choice can depend on what type of application you are using and what other libraries you are using.
In this mode the Event Manager registers
all active sockets and passes them to a select() call which then
processes the registered sockets. When select() returns the Event
Manager dispatches the appropriate handlers as the sockets get ready.
What to do:
Undefine WWW_WIN_ASYNC as a preprocessor directive when
generating the project for the Library. Please make sure that all
DLLs are compiled without this flag.
Here the Event Manager registers all
active sockets using WSAAsyncSelect which is part of the WinSock
API. When AsyncSelect() which is the asynchronous equivalent to
the select() returns, the Event Manager dispatches the results of
the AsyncSelect(). As the asynchronous select call needs a
windows handle, the Library creates a hidden window. This window is
not to be used explicitly by the application and the Library closes
the window when exiting.
What to do:
Define WWW_WIN_ASYNC as preprocessor directives when
generating the project for the Library.
A Windows application can use either a character based command line interface, or a graphic windows interface. Many MS-DOS applications do run as a simple command line tool which doesn't use GUI at all. Under Win32, there is a notion of a console application. This means that all user interaction happens through a standard DOS shell interface, with a FILE pointer like that in Unix. This does not exist under Win16, where a Windows window is required.
Libwww supports both the console and the windows interface, and again you must choose what version you prefer. This is often something you have to decide as you are creating the project.
If you want to make a windows application then this is the mode to use. Eric Prud'hommeaux has provided a Windows application wrapper for the W3C Line Mode Browser which you can use in order to build the browser as a Windows application. This is found in www.c. Three other modules, scroll.c and lib.c provide the window for the application.
What to do: Define
_WINDOWS as preprocessor directives when generating the project for the
Library.
The console option is available only in Win32 in which case all user interaction happens through the Win32 console window. This model strongly resembles a Unix vt100 terminal interface.
What to do: Define
_CONSOLE as preprocessor directives when generating the project for the
Library.
Windows has a concept of both static and dynamic libraries, the latter also known as DLLs. It is out of scope here to describe the difference between DLLs and static libraries, but as DLLs is based on a lot more flexible memory model it is almost always the best solution for Windows applications.
The W3C Sample Code Library support both models in that if can be built as either one big static library or as a set of small DLLs. As mentioned, it is in almost all cases recommended to build DLLs instead of static libraries, and on Win16 it is required because a static library is too big.
The libraries may be build as one large static library. This is how libwww is
implemented on Unix platforms. Subsequent references to the various DLLs may
all be assumed to refer to the staticly linked libwww. Care has been taken to
insure that there are no #define conflicts where one library
would want a #define that would interfere with the modules in
other libraries. When building a static library, see the following sections on
Select Options and Input/Output
Options, accumulate all the #defines that are required, and
build the whole of libwww with those #defines.
What to do:
Undefine WWW_WIN_DLL as preprocessor directives when
generating the project for the Library. Please note, that it is not
recommended to staticly link to the libraries if you are building a Win16
application as it creates segment size problems.
The libwww can also be built as a set of DLLs that follows the modular architecture of the Library. This enables the application programmer to choose exactly what functionality should be enabled in the application. The boundaries between these DLLs are based on module interdependency and some assumptions regarding which modules may be replaced by the application. Unlike static linking, dynamic linking requires that all the modules in a DLL be replaced at once. This is because the DLL needs all internal references to be resolved at build time.
What to do: Define
WWW_WIN_DLL as preprocessor directives when generating the project for
the Library.
In addition to platform independent modules, there is a small Windows specific
DLL which implements the trace message generation. The DLL is called
windll.dll and contains also the definition of the global trace
flag definition.
windll.c - DLLMain to all DLLs.
wwwdll.c - instances WWW_TraceFlag. This is included
in wwwdll so that it can export the variables to the rest of the DLLs.
wwwdll.def, wwwutils.def, wwwcore.def -
def (exports) files for the DLLs of the same names.
What to do: Include
the windll.dll as a part of your project and make sure that it is built as the
first DLL.
The functions exported from a DLL are listed in the EXPORTS
section of a .def file. These can be found in the windows directory. These may also be build by the
makeDefs.pl perl script, see the
description below. You can use the def files as a basis for generating the DLL
projects for your compiler.
What to do: Generate
the DLLs according to the def files so the exported interface is identical to
the set of functions defined in the actual c files included in the DLL.
Unfortunately, make files are not easily shared among different C compilers on Windows which complicates the distribution. See the installation instructions for details on possible ways of building libwww on windows.
This list gives an overview of the platforms where libwww is known (or belived) to work. Thanks a lot to the big group of creative contributors for the long list of hints, patches and additions to the make files and system dependent code! If you have any problems or questions then send a mail to the www-lib@w3.org mailing list (read the instructions for how to subscribe). We also keep an archive of the mailing list so that more easily can get up to speed.
| Platform | OS | Note |
|---|---|---|
| Unix | Tried on Solaris, Linux and Digital Unix | Build the library using GNU Autoconf's configure script as described in the installation instructions. Note: You must use GNU make in order to do this |
| Windows | Tried on Windows 95/98/NT | Follow the installation instructions and see the special notes for building on Windows |
| Macs | MacOS | It doesn't work out of the box but several people are using it on Macs. Ask on the <www-lib@w3.org> mailing list |
Disclaimer: Libwww has been designed to be both portable and flexible based on ANSI C and POSIX. However, there are platforms that do not conform to these standards. Even though autoconf is often capable of getting around such problems it may not always be the case. In this situation we are happy to accept patches but do not promise to maintain any platforms other than the ones we use for development.
This is a list of changes to libwww in reverse chronological order. See also the release notes for the latest release.
This is a minor release that mainly provides bug fixes and small enhancements that have been commited to CVS since the last release. Vladimir Koslov provided indispensable help for testing and preparing the Win32 release. Special thanks to Wayne Davison, Art Barstow, Peter Stamfest, Zhu Qun-Ying, Jens Meggers, Ken Olum for their contributions.
head.c, getheaders.c, chunk.c, chunkbody.c,
multichunk.c
When building libwww on Windows 95/98/NT, you can decide on a set of different options to be enabled or disabled:
Non-blocking network access can either be done using multiplexed I/O or
asynchronous I/O. Multiplexed I/O is based on an event loop with a
select() system call. The select() system call is a
function that scans a set of BSD like socket descriptors and returns when one
or more sockets are ready for either network read or write.
Asynchronous I/O is based on signals instead of a select() call.
Each time a socket is ready for a network operation, for example read,
a signal is generated. This signal can then be caught by the application in an
signal handler. Using the select() call is called
reactive whereas using signals is called pro-active.
Libwww can handle both approaches under Windows, and therefore you must choose which model you prefer. The choice can depend on what type of application you are using and what other libraries you are using.
In this mode the Event Manager registers
all active sockets and passes them to a select() call which then
processes the registered sockets. When select() returns the Event
Manager dispatches the appropriate handlers as the sockets get ready.
What to do:
Undefine WWW_WIN_ASYNC as a preprocessor directive when
generating the project for the Library. Please make sure that all
DLLs are compiled without this flag.
Here the Event Manager registers all
active sockets using WSAAsyncSelect which is part of the WinSock
API. When AsyncSelect() which is the asynchronous equivalent to
the select() returns, the Event Manager dispatches the results of
the AsyncSelect(). As the asynchronous select call needs a
windows handle, the Library creates a hidden window. This window is
not to be used explicitly by the application and the Library closes
the window when exiting.
What to do:
Define WWW_WIN_ASYNC as preprocessor directives when
generating the project for the Library.
A Windows application can use either a character based command line interface, or a graphic windows interface. Many MS-DOS applications do run as a simple command line tool which doesn't use GUI at all. Under Win32, there is a notion of a console application. This means that all user interaction happens through a standard DOS shell interface, with a FILE pointer like that in Unix. This does not exist under Win16, where a Windows window is required.
Libwww supports both the console and the windows interface, and again you must choose what version you prefer. This is often something you have to decide as you are creating the project.
If you want to make a windows application then this is the mode to use. Eric Prud'hommeaux has provided a Windows application wrapper for the W3C Line Mode Browser which you can use in order to build the browser as a Windows application. This is found in www.c. Three other modules, scroll.c and lib.c provide the window for the application.
What to do: Define
_WINDOWS as preprocessor directives when generating the project for the
Library.
The console option is available only in Win32 in which case all user interaction happens through the Win32 console window. This model strongly resembles a Unix vt100 terminal interface.
What to do: Define
_CONSOLE as preprocessor directives when generating the project for the
Library.
Windows has a concept of both static and dynamic libraries, the latter also known as DLLs. It is out of scope here to describe the difference between DLLs and static libraries, but as DLLs is based on a lot more flexible memory model it is almost always the best solution for Windows applications.
The W3C Sample Code Library support both models in that if can be built as either one big static library or as a set of small DLLs. As mentioned, it is in almost all cases recommended to build DLLs instead of static libraries, and on Win16 it is required because a static library is too big.
The libraries may be build as one large static library. This is how libwww is
implemented on Unix platforms. Subsequent references to the various DLLs may
all be assumed to refer to the staticly linked libwww. Care has been taken to
insure that there are no #define conflicts where one library
would want a #define that would interfere with the modules in
other libraries. When building a static library, see the following sections on
Select Options and Input/Output
Options, accumulate all the #defines that are required, and
build the whole of libwww with those #defines.
What to do:
Undefine WWW_WIN_DLL as preprocessor directives when
generating the project for the Library. Please note, that it is not
recommended to staticly link to the libraries if you are building a Win16
application as it creates segment size problems.
The libwww can also be built as a set of DLLs that follows the modular architecture of the Library. This enables the application programmer to choose exactly what functionality should be enabled in the application. The boundaries between these DLLs are based on module interdependency and some assumptions regarding which modules may be replaced by the application. Unlike static linking, dynamic linking requires that all the modules in a DLL be replaced at once. This is because the DLL needs all internal references to be resolved at build time.
What to do: Define
WWW_WIN_DLL as preprocessor directives when generating the project for
the Library.
In addition to platform independent modules, there is a small Windows specific
DLL which implements the trace message generation. The DLL is called
windll.dll and contains also the definition of the global trace
flag definition.
windll.c - DLLMain to all DLLs.
wwwdll.c - instances WWW_TraceFlag. This is included
in wwwdll so that it can export the variables to the rest of the DLLs.
wwwdll.def, wwwutils.def, wwwcore.def -
def (exports) files for the DLLs of the same names.
What to do: Include
the windll.dll as a part of your project and make sure that it is built as the
first DLL.
The functions exported from a DLL are listed in the EXPORTS
section of a .def file. These can be found in the windows directory. These may also be build by the
makeDefs.pl perl script, see the
description below. You can use the def files as a basis for generating the DLL
projects for your compiler.
What to do: Generate
the DLLs according to the def files so the exported interface is identical to
the set of functions defined in the actual c files included in the DLL.
Unfortunately, make files are not easily shared among different C compilers on Windows which complicates the distribution. See the installation instructions for details on possible ways of building libwww on windows.
This list gives an overview of the platforms where libwww is known (or belived) to work. Thanks a lot to the big group of creative contributors for the long list of hints, patches and additions to the make files and system dependent code! If you have any problems or questions then send a mail to the www-lib@w3.org mailing list (read the instructions for how to subscribe). We also keep an archive of the mailing list so that more easily can get up to speed.
| Platform | OS | Note |
|---|---|---|
| Unix | Tried on Solaris, Linux and Digital Unix | Build the library using GNU Autoconf's configure script as described in the installation instructions. Note: You must use GNU make in order to do this |
| Windows | Tried on Windows 95/98/NT | Follow the installation instructions and see the special notes for building on Windows |
| Macs | MacOS | It doesn't work out of the box but several people are using it on Macs. Ask on the <www-lib@w3.org> mailing list |
Disclaimer: Libwww has been designed to be both portable and flexible based on ANSI C and POSIX. However, there are platforms that do not conform to these standards. Even though autoconf is often capable of getting around such problems it may not always be the case. In this situation we are happy to accept patches but do not promise to maintain any platforms other than the ones we use for development.
This is a list of changes to libwww in reverse chronological order. See also the release notes for the latest release.
This is a minor release that mainly provides bug fixes and small enhancements that have been commited to CVS since the last release. Vladimir Koslov provided indispensable help for testing and preparing the Win32 release. Special thanks to Wayne Davison, Art Barstow, Peter Stamfest, Zhu Qun-Ying, Jens Meggers, Ken Olum for their contributions.
head.c, getheaders.c, chunk.c, chunkbody.c,
multichunk.c
When building libwww on Windows 95/98/NT, you can decide on a set of different options to be enabled or disabled:
Non-blocking network access can either be done using multiplexed I/O or
asynchronous I/O. Multiplexed I/O is based on an event loop with a
select() system call. The select() system call is a
function that scans a set of BSD like socket descriptors and returns when one
or more sockets are ready for either network read or write.
Asynchronous I/O is based on signals instead of a select() call.
Each time a socket is ready for a network operation, for example read,
a signal is generated. This signal can then be caught by the application in an
signal handler. Using the select() call is called
reactive whereas using signals is called pro-active.
Libwww can handle both approaches under Windows, and therefore you must choose which model you prefer. The choice can depend on what type of application you are using and what other libraries you are using.
In this mode the Event Manager registers
all active sockets and passes them to a select() call which then
processes the registered sockets. When select() returns the Event
Manager dispatches the appropriate handlers as the sockets get ready.
What to do:
Undefine WWW_WIN_ASYNC as a preprocessor directive when
generating the project for the Library. Please make sure that all
DLLs are compiled without this flag.
Here the Event Manager registers all
active sockets using WSAAsyncSelect which is part of the WinSock
API. When AsyncSelect() which is the asynchronous equivalent to
the select() returns, the Event Manager dispatches the results of
the AsyncSelect(). As the asynchronous select call needs a
windows handle, the Library creates a hidden window. This window is
not to be used explicitly by the application and the Library closes
the window when exiting.
What to do:
Define WWW_WIN_ASYNC as preprocessor directives when
generating the project for the Library.
A Windows application can use either a character based command line interface, or a graphic windows interface. Many MS-DOS applications do run as a simple command line tool which doesn't use GUI at all. Under Win32, there is a notion of a console application. This means that all user interaction happens through a standard DOS shell interface, with a FILE pointer like that in Unix. This does not exist under Win16, where a Windows window is required.
Libwww supports both the console and the windows interface, and again you must choose what version you prefer. This is often something you have to decide as you are creating the project.
If you want to make a windows application then this is the mode to use. Eric Prud'hommeaux has provided a Windows application wrapper for the W3C Line Mode Browser which you can use in order to build the browser as a Windows application. This is found in www.c. Three other modules, scroll.c and lib.c provide the window for the application.
What to do: Define
_WINDOWS as preprocessor directives when generating the project for the
Library.
The console option is available only in Win32 in which case all user interaction happens through the Win32 console window. This model strongly resembles a Unix vt100 terminal interface.
What to do: Define
_CONSOLE as preprocessor directives when generating the project for the
Library.
Windows has a concept of both static and dynamic libraries, the latter also known as DLLs. It is out of scope here to describe the difference between DLLs and static libraries, but as DLLs is based on a lot more flexible memory model it is almost always the best solution for Windows applications.
The W3C Sample Code Library support both models in that if can be built as either one big static library or as a set of small DLLs. As mentioned, it is in almost all cases recommended to build DLLs instead of static libraries, and on Win16 it is required because a static library is too big.
The libraries may be build as one large static library. This is how libwww is
implemented on Unix platforms. Subsequent references to the various DLLs may
all be assumed to refer to the staticly linked libwww. Care has been taken to
insure that there are no #define conflicts where one library
would want a #define that would interfere with the modules in
other libraries. When building a static library, see the following sections on
Select Options and Input/Output
Options, accumulate all the #defines that are required, and
build the whole of libwww with those #defines.
What to do:
Undefine WWW_WIN_DLL as preprocessor directives when
generating the project for the Library. Please note, that it is not
recommended to staticly link to