Journal Journal: Wanting it to break down
Documenting error messages. Most people would add it in there with, "No, I cannot make it. I have some belly button lint to sort, some paint I'm watching peel, and a few thousand error messages to document." Initially, I aimed to get a standard comment format for which I could write a little parser. Then she wrote a shell script of sed and awk tidbits to process the error message IDs.
So I wrote a parser for the output from *her* script. It might have been easier to maintain a parser just for the comments plus the subsequent log API calls, from which I could extract the error message IDs and format strings. But it would also mean I'd have almost no reason to go back and ask her repeatedly for help.
I did not plan it that way at all. Sometimes the world can be a wonderful place, however. Two smiles today.
She seemed to have lots of bugs to fix, so I could not figure out a good reason for it to take much time. In many ways, if we had time we could take it, because from an architectural point of view we really ought to scrap the entire existing log API in favor of a single, localizable file, build mechanics to generate unique IDs, and a simple parser to read the single file.
The work sounds terribly boring, but I wouldn't mind under certain circumstances. I'd just be waiting for the bugs to break it down again.