Lunchtime Reflections

[mdlbear]

Well, here I am at work, finishing lunch (I eat at my desk to leave time for a walk) and wondering whether to pick up a 300GB Maxtor drive at Fry's. They're only $80 this weekend. Since you were wondering, or even if you weren't, the price of a byte of rotating memory has been dropping steadily by a factor of two every year for the last half-century. That's roughly a factor of a quadrillion. If somebody offers you storage as a long-term investment, don't take it.

Now that I've been handed the expense check for last week's OSCON trip, I'm not feeling nearly as impoverished as I was this morning.

Here I am, at least two decades since research conclusively proved that flowcharts were worse than useless as a tool for software design, dusting off a copy of dia so that I can draw some flowcharts. For a software patent, no less. I'm going to need a shower when I get home.

Comments

[technoshaman.livejournal.com]

Flowcharts aren't, but I was always taught if you can't draw the picture from a data perspective, you probably don't understand what you're doing.

That and if you can't express it in two screenfuls (48 lines), break it out into a function.

Yay for new storage! What're the WD's running? I like those better, having run through a lot of them at $previous_work...

[patoadam]

Gee, I've been using flowcharts throughout my entire career. Perhaps because the first programming language I used professionally was a flowcharting language. I would draw the flowcharts and hand them to a coder, who would write the equivalent assembly language.

Ah, how software technology has changed!

Could you point me to some research on the uselessness of flowcharts?

[penngwyn.livejournal.com]

Strictly anecdotal, but I recall reading that most organizations that insisted on flow charts as part of module documentation wound up buying software that would draw the charts by reading the source code.

[mdlbear]

One of the first programs I wrote, at my first job in the mid-1960's, was a flowcharter. It was written in, and wrote flowcharts for, FORTRAN IV.

[mdlbear]

I'm too lazy to look it up, but if I remember correctly it was published in CACM roughly 20 years ago (give or take half a decade or so). Basically they had two groups of experienced programmers write programs -- one group started with flowcharts, the other group didn't. The group without flowcharts finished faster and had fewer bugs.

[patoadam]

At least the documentation matches the code that way. Until they modify the program and don't re-generate the flowcharts.

[patoadam]

In the situation I described, the main advantage of using flowcharts was that they were a higher-level language than the actual code.

I must admit that nowadays, I use flowcharts rarely, usually in situations where I find that their visual gestalt clarifies complex control structures better than source code.

Chacun à son goût.