Standard library: the "general and popular" rule, aka #1
- Posted by CChris <christian.cuvier at agriculture.gou?.?r> Apr 28, 2008
- 576 views
I'll be blunt: Please remove it. If you check the discussion lists, both for ESL on SourceForge and here about some include files, most recently math.e, you will notice that the main reason why they stagnated and eventually died is the debate over what is generic and what is "specific" or "rare", hence to be kept out in application of the "general and popular" principle. If someone has the minimalist urge and doesn't use something, s/he will post and say in substance: Don't include this, I don't use it. Add a few like this, and you come up with only a handful of additions to the current Eu distribution. At this point, the purpose of the extension usually vanishes. Look at most standard libraries in other languages. Perhaps there isn't a single programmer in that language that ever used 80% of routines in the library. I'll get flamed, but this not a liability, but, to the contrary, an asset. This is how you get people to program, because they see some flesh they can already put in their program, not just bare bones. To put it otherwise, a stdlib is useful because one can find in it most everything that might be standard for some subgroup of users. Your rule calls for a greater common denominator. Eu is in its current shape largely because of this error, repeated over time. Let's learn: what works is the lesser common multiple instead. My own alternative proposal would be: generic and useful in some known area of programming. CChris