Time Catcher Part 3


Well, I've been using my program to keep track of my time on a few projects and have made a few additions since the alpha release I talked about in my last post. Just taking a quick look through my gitk for the repository and I'm seeing a few noticable improvements:

The pause command is essentially the same as the start command with the switch flag, except that you don't have to be switching to a task ( you can actually take a break from tasks! ). It's quickly becoming a fast favorite of mine.

The delete command was born out of my scrolling through the view -a output and wanting to have a better way of deleting the tasks then manually removing the sequence and information files from the .tc directory. I use it everynow and then, although out of habit I write rm sometimes, it's starting to make me think I should shortcut rm to be delete.

The fixes in the list are minor things, the task state being wrong when finishing a task (it would show in progress until I fixed it), mind you it was only wrong during the finishing of the task's final output, not when it was viewed later.

The info file fix was something a bit silly where I forgot to not write out the taskInfo field of the struct that I was using to store the name of the info file itself. This caused the info file to end up with a bunch of filepaths at the end of the file. Not particularly harmful to the file, and useful sometimes, but not desired behavior.

The task resolution name fix is to handle task names with spaces in them to be shown correctly when using the view --all command. Before I was only reading out one string from the line to use as the name, and this was causing some minor annoyances when I was trying to delete a task and wasn't using it's full name.

Finally, I was surprised all the warning flags and pedantry passed in my makefile didn't catch this, but within the delete command I do ask for user confirmation. Out of sleep-deprived silliness I used char instead of int for the result of getc(stdin) and anyone whose read K & R will tell you that EOF is an int because it simply has to be. So I wasn't actually able to check the value of the user input again the EOF properly. Luckily, my friend happened to be compiling timecatcher on his raspberry pi and caught the warning for me.

That last warning catch is only on the develop branch of the repository as I didn't have time to do a hotfix or release yet. But it will be included in my next release, which should have a feature I've been wanting to implement:

    tcatch resume

In the same way that pause pauses the current task, I want to be able to resume the most recent task that was paused or finished. Why? Because I start up a task and work on it, then get distracted for a moment and pause my task, but then after responding to an email or whatever got in the way of coding, I start up the task I was previously working on. In fact, I do this so often I think a command would be a great idea for this.

The other feature I want to implement is using gnuplot and the sequence files of all tasks to plot some typeo of time spent on different tasks over the course of some time, and see how much time you spent on each one. These types of small statistics are useful for someone like me who wants to know what bogs them down the most so they can focus on getting better at it.

The only other addition to timecatcher is that my other friend Garth has expressed some interest in collaborating in it with me, so who knows maybe it won't just be my name on the commit log soon!

Other Posts

comments powered by Disqus