Jilara (jilara) wrote,

  • Mood:
  • Music:

Overthinking the software...

The so-called "help" tools available from the company that absorbed us are not as sophisticated as those we had developed, and our mandate was to move to their way of doing things. All fine and good, but they don't support command-line-level help. At all. And have no plan to do so. While our customers expect to be able to, within our Tcl-based language, type "help" and the command-name and get a file explaining syntax and usage.

So, I've been trying, for the past couple weeks, to understand a mostly undocumented proprietary help system, to try to disentangle it from the GUI-based help and modify it to give us command line help in the newest release. This is not helped by the fact that none of R&D understands the help system, the designer didn't comment his code, and I don't have access to the Tcl-based portion (which the R&D folk don't fully understand, anyway--I understand about 80% of the system, which is about 75% more than anyone except CMG, who are about 20% behind me). (A funny moment happened yesterday when my co-worker Sylvia demanded to be given part of my current project to work on, since she was under-employed the last day or two. "Fine," I said cheerfully. "I'm trying to decypher the help system. How much about C coding and makefiles do you understand?" Ummmm... Needless to say, she didn't hop in and join me.)

Well, last night, I woke up in the middle of the night with a revelation. I had a workaround that didn't require running the makefile at all. A lot of arcane processes and files were thus bypassed. IF I understand how the Tcl portion works, sorta, (which came to me in the night), I don't think we need it. We can patch around the makefile strategy by using some splices in the build tree. I talked to CMG, and they think it could work. And the thing about it is it's so *simple.* It doesn't use the previous strategy at all. I'd been overthinking the whole thing. Rather than building on the past system, I needed to create a new one.

I suppose the test will be whether the next build, incorporating my new strategy, actually works. I'm crossing my fingers.
  • Post a new comment


    Anonymous comments are disabled in this journal

    default userpic

    Your reply will be screened