Hi, I'm trying to learn Ruby and Socket programming. So far I managed to find some tutorials about Ruby and Socket programming and have made really simple applications like a Time server sending the time to a client. Now I wanted to try something a little more advanced. I want to make a client that will connect to server and receive messages and at the same time send messages to the server when certain events happen like when there's a file in a certain directory or a record in the database. So far from the tutorials I've studied, the clients all just wait for a response from the server. From what I understand once I do a gets I'll just be waiting for a reponse and will put the rest of the state waiting on that response. Is it possible to do both reading and writing on the socket at the same time? I also did some digging on threads. I had the idea of making 2 threads one to receive messages from the server and one to check the directory/database then send a message. Again I don't see though how I can do that based on what I done in the tutorials. It seems like my threads will just hang when I try to read from the server. Is this the proper approach? Are there other methods aside from using threads? I saw topic on a multiplexing server but all the examples were just waiting again on a read from the client. I know these are pretty broad questions and topics but hoping someone can help clarify these issues and point me at the right direction. Thanks!
on 2008-11-04 11:50
on 2008-11-08 03:33
Threads can work, but most networking code I've written uses Kernel#select. It's an incredibly powerful, yet somewhat hard to master, method that allows you to read and write on many sockets (actually IO objects) at the same time. It sounds like you want two read objects, one for the server and one for the database, and that you want to write to the network socket (yes, can you can read and write to the same socket - just use "puts" as well as "gets", or whatever equivalent methods you like) in the part of the code that handles what happens when the database socket is ready to be read. http://www.rubycentral.com/book/ref_m_kernel.html#... HTH, Adam
on 2008-11-08 07:34
check out eventmachine :) -=R
on 2008-11-10 04:47
Adam B. wrote: > Threads can work, but most networking code I've written uses > Kernel#select. It's an incredibly powerful, yet somewhat hard to > master, method that allows you to read and write on many sockets > (actually IO objects) at the same time. It sounds like you want two > read objects, one for the server and one for the database, and that > you want to write to the network socket (yes, can you can read and > write to the same socket - just use "puts" as well as "gets", or > whatever equivalent methods you like) in the part of the code that > handles what happens when the database socket is ready to be read. > > http://www.rubycentral.com/book/ref_m_kernel.html#... > > HTH, > > Adam Thanks Adam! I looked for examples using select and I mostly found chat servers. I tried it out and it does look like an effective and powerful approach. I still though don't understand how to trigger a read from another socket if I want to monitor a directory or a database. I wish though I could find better examples in the internet. How would you set up a socket to monitor a directory or database or is my understanding of the matter off? I checked out EventMachine as well which looks great just that since I want to learn I want to try out making something from scratch.
on 2008-11-10 22:37
> How would you set up a socket to > monitor a directory or database or is my understanding of the matter > off? > Sockets are just a conduit for data. You can use their blocking capability (via select, etc) to control some portion of the interaction between the client and server (or whatever relationship you deem appropriate between the end points). Regardless, you still require an active process/thread to 'put' the data into one end of the socket and another to active process/thread to 'get' it at the other end. It's much easier if you maintain the separation of client/network/server activities in your mind and in the code. Another consideration is that this a glorious intersection of things (both threads and sockets) that are similar across multiple platforms but only similar. Each platform has it's own features (quirks). The source code for webrick and the net library in lib provide plenty of examples of how threads and sockets interact (or not) in an internet application. It's one of the beauties of Ruby. You get the code.
on 2008-11-13 05:54
Thanks Jeff! I'll check out the code and study them. Hopefully a novice like me can pick it up quickly.