Saturday, May 27, 2017

Building software tools

How much time one should spend building the software the customers want, versus how much time one should spend building tools to better enable one to build the software the customers want, is simply a special case of how "round-about" one should make any production process.

My friend Howard Baetjer noted this many years ago, but it seems it is still not widely recognized in the software industry.

5 comments:

  1. IMO there are too many "frameworks" and meta tools. I cite as evidence the continual emergence of new ones. The toaster you use is essentially unchanged from the 1920s.
    I have seen projects go years long and almost 200m dollars over budget because of the insistence of using such a framework.

    ReplyDelete
    Replies
    1. Well, we certainly need a balance! But remember C is a "meta tool" over assembler: we also don't want to forego all such tools!

      Delete
    2. Also, that toaster is certainly not *produced* using the same tools as in 1920! And:

      "I cite as evidence the continual emergence of new ones."

      Would you say the continual emergence of new factory production techniques means there are too many factory production techniques?

      Delete
    3. When they are produced and adopted faster than the training time required to master them, yes. C was a commonly shared tool. So was Java or XML.So is now Hadoop. But within say the Java ecosystem there is a huge proliferation of tools that I can with confidence my fortune 10 employer would have been better off without. Cannot hire people, constant fixes and versions, incompatibilities between them, etc.

      Delete
    4. Yes: for sure this can go overboard!

      Delete

Distraction Deterrents in Small Contexts

"distracted from distraction by distraction" - T.S. Eliot I've been reading a little on how Facebook and other social netwo...