The application was about 25KLOC with about 2MLOC of Perl classes that, essentially, were configuration information. I had several scripts that managed the generation and updates (which were frequent) to these classes that represented messages that the application I worked on needed to handle. This was in addition to any CPAN code.
My criteria for good software:
- Does it work?
- Can someone else come in, make a change, and be reasonably certain no bugs were introduced?
| [reply] |
But if the "config classes" aren't the single source of this information and you didn't treat it as code -- is it source code? Or is it something else?
/J
| [reply] |
These were the classes that implemented both the messages and the components of the messages. I certainly treated them as code within the application, I just didn't physically type the code for them.
The management part came in from the fact that my application was a simulator used to test another section of the overall network. I had to support 3 revisions of the messages, plus variations of the messages depending on what's been implemented on the section my app tested. So, at any given time, I might have 5-7 different variations of a given message in play at a given time. I probably invested 2-3KLOC in the management scripts, part of which was parsing the original spec for the messages (in XLS, if I remember correctly).
My criteria for good software:
- Does it work?
- Can someone else come in, make a change, and be reasonably certain no bugs were introduced?
| [reply] |