23 skills found
PaddyCahil / Windows Api Function CheatsheetsA reference of Windows API function calls, including functions for file operations, process management, memory management, thread management, dynamic-link library (DLL) management, synchronization, interprocess communication, Unicode string manipulation, error handling, Winsock networking operations, and registry operations.
CharlesJS / SwiftyXPCA wrapper for Apple’s XPC interprocess communication library that gives it a type-safe, easy-to-use interface utilizing Codable and Swift Concurrency.
0x1eef / Xchan.rbAn easy to use InterProcess Communication (IPC) library.
SOYJUN / TCP Socket Client ServerThe aim of this assignment is to have you do TCP socket client / server programming using I/O multiplexing, child processes and threads. It also aims at getting you to familiarize yourselves with the inetd superserver daemon, the ‘exec’ family of functions, various socket error scenarios, some socket options, and some basic domain name / IP address conversion functions. Apart from the material in Chapters 1 to 6 covered in class, you will also need to refer to the following : the exec family of functions (Section 4.7 of Chapter 4) using pipes for interprocess communication (IPC) in Unix error scenarios induced by process terminations & host crashes (Sections 5.11 to 5.16, Chapter 5) setsockopt function & SO_REUSEADDR socket option (Section 7.2 & pp.210-213, Chapter 7) gethostbyname & gethostbyaddr functions (Sections 11.3 & 11.4, Chapter 11) the basic structure of inetd (Section 13.5, Chapter 13) programming with threads (Sections 26.1 to 26.5, Chapter 26) Overview I shall present an overview of this assignment and discuss some of the specification details given below in class on Wednesday, September 17 & Monday, September 22. Client The client is evoked with a command line argument giving either the server IP address in dotted decimal notation, or the server domain name. The client has to be able to handle either mode and figure out which of the two is being passed to it. If it is given the IP address, it calls the gethostbyaddr function to get the domain name, which it then prints out to the user in the form of an appropriate message (e.g., ‘The server host is compserv1.cs.stonybrook.edu’). The function gethostbyname, on the other hand, returns the IP address that corresponds to a given domain name. The client then enters an infinite loop in which it queries the user which service is being requested. There are two options : echo and time (note that time is a slightly modified version of the daytime service – see below). The client then forks off a child. After the child is forked off, the parent process enters a second loop in which it continually reads and prints out status messages received from the child via a half-duplex pipe (see below). The parent exits the second loop when the child closes the pipe (how does the parent detect this?), and/or the SIGCHLD signal is generated when the child terminates. The parent then repeats the outer loop, querying the user again for the (next) service s/he desires. This cycle continues till the user responds to a query with quit rather than echo or time. The child process is the one which handles the actual service for the user. It execs (see Section 4.7, Chapter 4) an xterm to generate a separate window through which all interactions with server and user take place. For example, the following exec function call evokes an xterm, and gets the xterm to execute echocli, located in the current directory, passing the string 127.0.0.1 (assumed to be the IP address of the server) as the command line argument argv[1] to echocli (click on the url for further details) : execlp("xterm", "xterm", "-e", "./echocli", "127.0.0.1", (char *) 0) xterm executes one of two client programs (echocli or timecli, say) depending on the service requested. A client program establishes a TCP connection to the server at the ‘well-known port’ for the service (in reality, this port will, of course, be some ephemeral port of your choosing, the value of which is known to both server and client code). All interaction with the user, on the one hand, and with the server, on the other, takes place through the child’s xterm window, not the parent’s window. On the other hand, the child will use a half-duplex pipe to relay status information to the parent which the parent prints out in its window (see below).To terminate the echo client, the user can type in ^D (CTRL D, the EOF character). To terminate the time client, the only option is for the user to type in ^C (CTRL C). (This can also be used as an alternative means of terminating the echo client.) Note that using ^C in the context of the time service will give the server process the impression that the client process has ‘crashed’. It is your responsibility to ensure that the server process handles this correctly and closes cleanly. I shall address this further when discussing the server process. It is also part of your responsibility in this assignment to ensure that the client code is robust with respect to the server process crashing (see Sections 5.12 & 5.13, Chapter 5). Amongst other implications, this means that it would probably be a good idea for you to implement your echo client code along the lines of either : Figure 6.9, p.168 (or even Figure 6.13, p.174) which uses I/O multiplexing with the select function; or of Figure 26.2, p.680, which uses threads; rather than along the lines of Figure 5.5, p.125. When the child terminates, either normally or abnormally, its xterm window disappears instantaneously. Consequently, any status information that the child might want to communicate to the user should not be printed out on the child’s xterm window, since the user will not have time to see the final such message before the window disappears. Instead, as the parent forks off the child at the beginning, a half-duplex pipe should be established from child to parent. The child uses the pipe to send status reports to the parent, which the parent prints out in its window. I leave it up to you to decide what status information exactly should be relayed to the parent but, at a minimum, the parent should certainly be notified, in as precise terms as possible, of any abnormal termination conditions of the service provided by the child. In general, you should try to make your code as robust as possible with respect to handling errors, including confused behaviour by the user (e.g., passing an invalid command line argument; responding to a query incorrectly; trying to interact with the service through the parent process window, not the child process xterm; etc.). Amongst other things, you have to worry about EINTR errors occurring during slow system calls (such as the parent reading from the pipe, or, possibly, printing to stdout, for example) due to a SIGCHLD signal. What about other kinds of errors? Which ones can occur? How should you handle them? Server The server has to be able to handle multiple clients using threads (specifically, detached threads), not child processes (see Sections 26.1 to 26.4, Chapter 26). Furthermore, it has to be able to handle multiple types of service; in our case, two : echo and time. echo is just the standard echo service we have seen in class. time is a slightly modified version of the daytime service (see Figure 1.9, p.14) : instead of sending the client the ‘daytime’ just once and closing, the service sits in an infinite loop, sending the ‘daytime’, sleeping for 5 seconds, and repeating, ad infinitum. The server is loosely based on the way the inetd daemon works : see Figure 13.7, p.374. However, note that the differences between inetd and our server are probably more significant than the similarities: inetd forks off children, whereas our server uses threads; inetd child processes issue exec commands, which our server threads do not; etc. So you should treat Figure 13.7 (and Section 13.5, Chapter 13, generally) as a source of ideas, not as a set of specifications which you must slavishly adhere to and copy. Note, by the way, that there are some similarities between our client and inetd (primarily, forking off children which issue execs), which could be a useful source of ideas. The server creates a listening socket for each type of service that it handles, bound to the ‘well-known port’ for that service. It then uses select to await clients (Chapter 6; or, if you prefer, poll; note that pselect is not supported in Solaris 2.10). The socket on which a client connects identifies the service the client is seeking. The server accepts the connection and creates a thread which provides the service. The thread detaches itself. Meanwhile, the main thread goes back to the select to await further clients. A major concern when using threads is to make sure that operations are thread safe (see p.685 and on into Section 26.5). In this respect, Stevens’ readline function (in Stevens’ file unpv13e/lib/readline.c, see Figure 3.18, pp.91-92) poses a particular problem. On p.686, the authors give three options for dealing with this. The third option is too inefficient and should be discarded. You can implement the second option if you wish. Easiest of all would be the first option, since it involves using a thread-safe version of readline (see Figures 26.11 & 26.12) provided in file unpv13e/threads/readline.c. Whatever you do, remember that Stevens’ library, libunp.a, contains the non-thread-safe version of Figure 3.18, and that is the version that will be link-loaded to your code unless you undertake explicit steps to ensure this does not happen (libunp.a also contains the ‘wrapper’ function Readline, whose code is also in file unpv13e/lib/readline.c). Remaking your copy of libunp.a with the ‘correct’ version of readline is not a viable option because when you hand in your code, it will be compiled and link-loaded with respect to the version of libunp.a in the course account, ~cse533/Stevens/unpv13e_solaris2.10 (I do not intend to change that version since it risks creating confusion later on in the course). Also, you will probably want to use the original version of readline in the client code anyway. I am providing you with a sample Makefile which picks up the thread-safe version of readline from directory ~cse533/Stevens/unpv13e_solaris2.10/threads and uses it when making the executable for the server, but leaves the other executables it makes to link-load the non-thread-safe version from libunp.a. Again, it is part of your responsibility to make sure that your server code is as robust as possible with respect to errors, and that the server threads terminate cleanly under all circumstances. Recall, first of all, that the client user will often use ^C (CTRL C) in the xterm to terminate the service. This will appear to the server thread as if the client process has crashed. You need to think about the error conditions that will be induced (see Sections 5.11 to 5.13, Chapter 5), and how the echo and time server code is to detect and handle these conditions. For example, the time server will almost certainly experience an EPIPE error (see Section 5.13). How should the associated SIGPIPE signal be handled? Be aware that when we return out of the Stevens’ writen function with -1 (indicating an error) and check errno, errno is sometimes equal to 0, not EPIPE (value 32). This can happen under Solaris 2.10, but I am not sure under precisely what conditions nor why. Nor am I sure if it also happens under other Unix versions, or if it also happens when using write rather than writen. The point is, you cannot depend on errno to find out what has happened to the write or writen functions. My suggestion, therefore, is that the time server should use the select function. On the one hand, select’s timeout mechanism can be used to make the server sleep for the 5 seconds. On the other hand, select should also monitor the connection socket read event because, when the client xterm is ^C’ed, a FIN will be sent to the server TCP, which will prime the socket for reading; a read on the socket will then return with value 0 (see Figure 14.3, p. 385 as an example). But what about errors other than EPIPE? Which ones can occur? How should you handle them? Recall, as well, that if a thread terminates without explicitly closing the connection socket it has been using, the connection socket will remain existent until the server process itself dies (why?). Since the server process is supposed, in principle, to run for ever, you risk ending up with an ever increasing number of unused, ‘orphaned’ sockets unless you are careful. Whenever a server thread detects the termination of its client, it should print out a message giving appropriate details: e.g., “Client termination: EPIPE error detected”, “Client termination: socket read returned with value 0”, “Client termination: socket read returned with value -1, errno = . . .”, and so on. When debugging your server code, you will probably find that restarting the server very shortly after it was last running will give you trouble when it comes to bind to its ‘well-known ports’. This is because, when the server side initiates connection termination (which is what will happen if the server process crashes; or if you kill it first, before killing the client) TCP keeps the connections open in the TIME_WAIT state for 2MSLs (Sections 2.6 & 2.7, Chapter 2). This could very quickly become a major irritant. I suggest you explore the possibility of using the SO_REUSEADDR socket option (pp.210-213, Chapter 7; note that the SO_REUSEPORT socket option is not supported in Solaris 2.10), which should help keep the stress level down. You will need to use the setsockopt function (Section 7.2) to enable this option. Figure 8.24, p.263, shows an instance of server code that sets the SO_REUSEADDR socket option. Finally, you should be aware of the sort of problem, described in Section 16.6, pp.461-463, that might occur when (blocking) listening sockets are monitored using select. Such sockets should be made nonblocking, which requires use of the fcntl function after socket creates the socket, but before listen turns the socket into a listening socket.
nmittu / IpcSimple interprocess communication library with c, c++, nodejs and Java API'S
domfarolino / Mage🧙♂️ A simple interprocess communication (IPC) library
PCWeChat / Mmmojommmojo is a library for interprocess communication (IPC) based on chromium mojo.
bb107 / ServiceCallInterprocess communication library, providing the ability to call functions from each other
keithrausch / InterprocessMemPoolc++ library for interprocess memory pools, communication, and automatic network device discovery. lightweight DDS alternative.
svebert / InterprocessMsgLightweight library for bidirectional communication between two processes on the same computer. The library is cross-platform compatible, tested with Windows 10 on Intel processer and Ubuntu Mate on a ARM processor. This lib is based on boost interprocess
jashwanth / Remote Code PublisherRemote-Code-Publisher Purpose: A Code Repository is a Program responsible for managing source code resources, e.g., files and documents. A fully developed Repository will support file persistance, managment of versions, and the acquisition and publication of source and document files. A Remote Repository adds the capability to access the Repository's functionality over a communication channel, e.g., interprocess communication, inter-network communication, and communication across the internet. In this project we will focus on the publication functionality of a Remote Repository. We will develop a remote code publisher, local client, and communication channel that supports client access to the publisher from any internet enabled processor. The communication channel will use sockets and support an HTTP like message structure. The channel will support: HTTP style request/response transactions One-way communication, allowing asynchronous messaging between any two endpoints that are capable of listening for connection requests and connecting to a remote listener. Transmission of byte streams that are set up with one or more negotiation messages followed by transmission of a stream of bytes of specified stream size2. The Remote Code Publisher will: Support publishing web pages that are small wrappers around C++ source code files, just as we did in Project #3. Accept source code text files, sent from a local client. Support building dependency relationships between code files saved in specific repository folders, based on the functionality you provided in Project #2 and used in Project #3. Support HTML file creation for all the files in a specified repository folder1, including linking information that displays dependency relationships, and supports and navigation based on dependency relationships. Delete stored files, as requested by a local client. Clients of the Remote Code Publisher will provide a Graphical User Interface (GUI) with means to: Upload one or more source code text files to the Remote Publisher, specifying a category with which those files are associated1. Display file categories, based on the directory structure supported by the Repository. Display all the files in any category. Display all of the files in any category that have no parents. Display the web page for any file in that file list by clicking within a GUI control. This implies that the client will download the appropriate webpages, scripts, and style sheets and display, by starting a browser with a file cited on the command line2. On starting, will download style sheet and JavaScript files from the Repository. Note that your client does not need to supply the functionality to display web pages. It simply starts a browser to do that. Browsers will accept a file name, which probably includes a relative path to display a web page from the local directory. You could also start IIS web server and provide an appropriate URL to the browser on startup. Either approach is acceptable. If you use IIS, you won't have to download files, but you are obligated to show that you can do that. Requirements: Your Remote Repository: (2) Shall use Visual Studio 2015 and its C++ Windows console projects, as provided in the ECS computer labs. You must also use Windows Presentation Foundation (WPF) to provide a required client Graphical User Interface (GUI). (1) Shall use the C++ standard library's streams for all console I/O and new and delete for all heap-based memory management. (3) Shall provide a Repository program that provides functionality to publish, as linked web pages, the contents of a set of C++ source code files. (4) Shall, for the publishing process, satisfy the requirements of CodePublisher developed in Project #3. (4) Shall provide a Client program that can upload files3, and view Repository contents, as described in the Purpose section, above. (3) Shall provide a message-passing communication system, based on Sockets, used to access the Repository's functionality from another process or machine. (2) The communication system shall provide support for passing HTTP style messages using either synchronous request/response or asynchronous one-way messaging. (1) The communication system shall also support sending and receiving streams of bytes6. Streams will be established with an initial exchange of messages. (5) Shall include an automated unit test suite that demonstrates you meet all the requirements of this project4 including the transmission of files. (5 point bonus) Shall optionally use a lazy download strategy, that, when presented with a name of a source code web page, will download that file and all the files it links to. This allows you to demonstrate your project using local webpages instead of downloading the entire contents of the Code Publisher for demonstration. (5 point bonus) Shall optionally have the publisher accept a path, on the commandline, to a virtual directory on the server. Then support browsing directly from the server by supplying a url to that path when you start a browser. This works only if you setup IIS on your machine and make the path a virtual directory. The TAs will do that on the grading machines. Categories are the names of folders in which the Repository stores its source code and web files. You may define Categories in any way that seems sensible. For example, they could simply be the namespace(s) for the uploaded files, or a Client supplied name. You will find a demonstration of how to programmatically start an application here. The stream capablity is intended to send files, which could be either text or binary format. Stream size will be the file size. Transmitting and receiving byte streams will be used to send and receive files in either text or binary format. This is in addition to the construction tests you include as part of every package you submit. Project 3 statement: Purpose: A Code Repository is a Program responsible for managing source code resources, e.g., files and documents. A fully developed Repository will support file persistance, managment of versions, and the acquisition and publication of source and document files. This project focuses on just the publishing functionality of a repository. In this project we will develop means to display source code files as web pages with embedded child links. Each link refers to a code file that the displayed code file depends on. There are several things you need to know in order to complete this project: Each file to be published is a C++ source file. Our publisher will generate, for each of these, an HTML file, with most of the contents drawn from the code file. The pages we will generate have only static content, with the exception of some embedded JavaScript and styling, so we won't need a web server. We will need to preserve the white space structure of the displayed source code. That can be done embedding all the code between the tags <pre> and </pre> or by using the CSS white-space property with value "pre" to style a div with all the code in its contents. Any markup characters in the code text will have to be escaped, e.g., replace < with < and > with >. File dependencies are displayed in the web page with embedded links, which are implemented in HTML5 with anchor elements: <a href="[url of referenced html page]">source code file name</a> For each class, we will, optionally, implement outlining, similar to the visual studio outlining feature. To do that we will use the CSS display property, with values: normal or none, to control whether the contents of a div are visible or not. The Code Publisher will be embedded in a mock Repository with almost no functionality except to support publishing of source code as web pages. Specifically you are not expected to provide support for: package checkin or checkout versioning You are expected to support: Dependency analysis of the C++ source code files you will publish, using the analyzer you developed in Project #2. The ability to specify, on the command line, files to be published, by providing command line arguments for path and file patterns. The ability to display any file cited on the command line, by starting a process that runs a browser of your choice, naming the specification of the file you want to display. Note that the CodePublisher project creates a code generator. Its inputs are C++ code and its outputs are HTML code. Requirements: Your CodePublisher Project: (1) Shall use Visual Studio 2015 and its C++ Windows console projects, as provided in the ECS computer labs. (2) Shall use the C++ standard library's streams for all console I/O and new and delete for all heap-based memory management1. (4) Shall provide a Publisher program that provides for creation of web pages each of which captures the content of a single C++ source code file, e.g., *.h or *.cpp. (10) Shall, optionally2 provide the facility to expand or collapse class bodies, methods, and global functions using JavaScript and CSS properties. (2) Shall provide a CSS style sheet that the Publisher uses to style its generated pages and (if you are implementing the previous optional requirement) a JavaScript file that provides functionality to hide and unhide sections of code for outlining, using mouse clicks. (2) Shall embed in each web page's <head> section links to the style sheet and JavaScript file. (4) Shall embedd HTML5 links to dependent files with a label, at the top of the web page. Publisher shall use functionality from your Project #2 to discover package dependencies within the published set of source files. (2) Shall develop command line processing to define the files to publish by specifying path and file patterns. (3) Shall demonstrate the CodePublisher functionality by publishing all the important packages in your Project #3. (5) Shall include an automated unit test suite that demonstrates you meet all the requirements of this project2. That means that you are not allowed to use any of the C language I/0, e.g., printf, scanf, etc, nor the C memory management, e.g., calloc, malloc, or free. This optional requirement will take a significant amount of work to complete successfully. You should get everything else working before attempting this additional effort. This is in addition to the construction tests you include as part of every package you submit.
AntonyMei / SharedMemoryObjectStoreA high throughput low latency shared memory object store specially optimized for RL.
alekshura / Compentio.Pipes.NET 5 Named Pipes interprocess communication library
kata198 / ShmfileSmall shared library to use shared memory as a FILE* stream , with interprocess communication in mind
affix / Windows Api Function CheatsheetsA comprehensive reference of Windows system calls, including functions for file operations, process management, memory management, thread management, dynamic-link library (DLL) management, synchronization, interprocess communication, Unicode string manipulation, error handling, Winsock networking operations, and registry operations.
clayv / FastIpcA very fast interprocess communication library using memory mapped files.
DemiMarie / SlipRockA secure local interprocess communication library.
hecrj / PlugA library for type-safe interprocess communication in Rust
omikhailov / CorpusCallosumLibrary for interprocess communication between noncontainerized .Net and containerized UWP apps and services
cdalizadeh / SimpleIpcOpen source .NET library for interprocess communication over TCP
