Backup Now, Please.

I realize it's been over a year since my last post, and I have a bunch of stuff stacked up. But this was pretty slick and frankly, I don't want to lose it. So here goes.

The problem: Oftentimes when I'm working on a project, being the hard-drive-paranoid-soul that I am, as I finish a milestone I'll run a Time Machine backup, and also sometimes rsync my project directory to Dropbox. However, it annoys me to shift to the mouse/trackpad to run the time-machine backup, then switch to the Terminal to run the rsync backup.

The solution: A decidedly love-sided instance of my love-hate relationship with AppleScript

do shell script "/System/Library/CoreServices/backupd.bundle/Contents/Resources/backupd-helper >/dev/null 2>&1 &"

property nullStr : ""
set defaultSrc to "/path/to/project/dir"
set defaultDest to "/path/to/destination/dir/in/Dropbox"

--- cancel actions from either of these dialogs will effectively cancel the script
set dropBoxSource to text returned of (display dialog "Dropbox backup source?" default answer defaultSrc)
set dropBoxDest to text returned of (display dialog "Dropbox backup destination?" default answer defaultDest)

if dropBoxSource is not equal to "" and dropBoxDest is not equal to "" then
set rsyncCmd to "rsync -rav " & dropBoxSource & " " & dropBoxDest
display dialog ("Run: '" & rsyncCmd & "'")
--- if the user cancels the dialog here, the rsync won't run. this is just what we want :)

--- set a 2-hour timeout for the rsync command
with timeout of 7200 seconds
 do shell script rsyncCmd
end timeout
end if

I saved this applescript as an application, which I can then invoke via spotlight/Quicksilver/Alfred etc, thus removing the problem of hunting for the Time Machine menu.

DISCLAIMER:

The shell script that runs the Time Machine backup is a non-documented, non-supported, run-at-your-own-risk solution. It works for me, but I'm making absolutely no guarantees that it won't break something. For that matter, you ought to know what rsync does before feeling too comfy with the rsync backup either.

Breaking long lines in Vim

Vim allows you to set a line length at which to break long lines, which I use in text files to break at 80 characters. However, in source files, I don't want it to break lines right at 80 characters lest it break in the middle of something that doesn't want to be broken in.

So I put the following in my .vimrc as a shortcut to breaking at 80 characters only when I want it to:


map \br 80\|? r


This tells Vim to go to column 80, search backward for the previous " " (space), and replace it with a carriage return.

I use this primarily for breaking comment lines at 80 characters. If Vim is set up properly (I can't remember the setting off the top of my head, but maybe I'll post it here later) the new line you inserted will be marked as a comment also (in the case of # or " comment markers, as this will obviously work with C/Java style multi-line comments).

Vim Sneakiness

One of the things that I like most about Xcode is the auto-insertion of a closing curly-brace on the second line below whenever I enter an opening curly-brace, and the insertion of the cursor on the line in-between them. However, at heart I'm still a Vim guy, and tonight I got tired of actually typing that closing curly-brace. So, to make Vim automagically create my curly-brace blocks for me, I did the following.

In my .vimrc, I place the following lines

if has("autocmd")
    autocmd FileType c source ~/.vim/brace.vim
    autocmd FileType cpp source ~/.vim/brace.vim
    autocmd FileType objc source ~/.vim/brace.vim
    autocmd FileType java source ~/.vim/brace.vim
    autocmd FileType pl source ~/.vim/brace.vim
endif

brace.vim is very simple, containing only the following line:

imap { {<CR>}<Esc>O

This causes typing a opening curly brace to immediately insert a Carriage Return character (equivalent to pressing return/enter) then inserts a closing curly-brace, then switching to command-mode and "typing" O to place an insert-mode cursor on a newly-inserted line above the closing curly-brace.

If anyone has a better way to do this, I'm open to suggestions.

Here's a handy little tool I've used often...

Problem: In several applications, I've found the need to pop up an alert/message window telling the application user that something went wrong, or that something they did wasn't kosher.

Solution: NSRunAlertPanel. Little macros like this one are additional reasons I love programming for the Mac. But to make it a little more useful, I created a method call that makes a dummy alert with an ok button and a given message. It looks like this:

-(void) displayAlert:(NSString*) alertMsg {
    NSRunAlertPanel(@"Alert", alertMsg, @"OK", nil, nil);
}

The actual NSRunAlertPanel macro signature looks like this:

NSRunAlertPanel(NSString *title,
            NSString *msgFormat,
            NSString *defaultButton,
            NSString *alternateButton,
            NSString *otherButton)

Other parameters can be added to the method signature to create more full-featured alerts. I created the method solely to remove the NSRunAlertPanel call (which can be a big method call) from the logic of what I'm doing at the moment and replace it with a short method call that does what I want it to.

Publishing a Bonjour networking service with NSNetService

So I'm working on an undisclosed (but totally awesome) iPhone/iPod Touch app for CS 598R (a capstone/senior project class), and have developed the following procedure for publishing a Bonjour service. I'm posting it here so I can always know where to find code that does what I want.

Feel free to use the code, just don't blame me if it doesn't do what you want it to...

You can see the basic sample code here, or on the BYU CocoaHeads site here.

NSMutableArray makes awesome Cocoa stacks and queues

A difference between Java and Objective-C/Cocoa? Java has ConcurrentLinkedQueue, PriorityBlockingQueue, ArrayBlockingQueue, blah blah blah.

Objective-C/Cocoa(or Foundation) has NSMutableArray. NSMutableArray has some nice instance methods that make it _extremely_ easy to build a queue, stack, priority queue (synchronized or otherwise) without loading up the API with a billion separate classes.

Need a stack?

-(void) push:(id) item {
[list addObject:item] // where list is the actual array in your stack
count++;
}

-(id) pop {
id r = [list lastObject];
[list removeLastObject];
count--;
return r;
}

How about a queue?


-(void) enqueue:(id) item {
[list insertObject:item atIndex:0];
count++;
}

-(id) dequeue {
id r = [list lastObject];
[list removeLastObject];
count--;
return r;
}

Priority queueing and/or synchronization for thread-safety will be left as an exercise for the reader. Man, I've always wanted to say that... :)

NSThreads when you need to wait.

For my webserver project I need to use queues. The main thread places connections that need to be serviced in a queue. "Worker" threads need to place log entries in a queue to be written out to the log file. These queues obviously need to be thread-safe. Writing a thread-safe queue is quite simple thanks to the NSLock and NSConditionLock classes. Just like any other semaphore, you create a NSLock object and surround your critical section with:

[myLock lock]
critical section...
[myLock unlock]

The real question I had to deal with here, though, was thoroughly testing my queues to make sure that they were _really_ thread safe. Unlike POSIX threads, Cocoa/Foundation has no mechanism for waiting on NSThread objects. But I needed to know that all of my testing threads were finished before I started comparing expected and actual results. The solution is not that difficult. The entry point/run loop for an NSThread object is a method with the mandatory signature:

(void) threadFuncName:(id)

So you create a struct or a class that has any data you need to pass in to your thread's main function. In addition, include in this struct or class an object of NSConditionLock. Immediately upon entry to the function, lock this with some condition code that will let your main program thread know that the thread is running. After spawning the thread, you want the main program execution to try and lock your condition lock when the run condition is no longer true. This will cause your main program thread to block and wait for your new thread to unlock with the condition you're waiting on. The last thing you'll want to do before your thread exits the run loop function is unlock the NSConditionLock with some condition that will let your main program thread know it's done. This will unblock your main thread of execution and you're off.
Sample code: (not complete, just the snippets that you'll need)

#import <Foundation/Foundation.h>
#define NOT_DONE 0
#define DONE 1

// this will need to be on some class that is calling the NSThread object
-(void) someFunc:(id) arg {
NSConditionLock* myLock = arg;
[myLock lock];
//do stuff that we need to know about in main
[myLock unlockWithCondition:DONE];
}


int main() {

NSConditionLock* finishedLock = [[NSConditionLock alloc]
initWithCondition: NOT_DONE];

NSThread* myThread = [[NSThread alloc] initWithTarget:self
selector:@selector(someFunc:) object:finishedLock];

[myThread start];

[finishedLock lockWhenCondition:DONE];

// now we can do whatever we need to do with the results from
// the thread we spawned

}