On Mon, Nov 03, 2003 at 10:43:28AM -0800, David Fox wrote: > The least important reason is size. We are creating lots of small > scripts, such as those you find in /etc/init.d, as part of our hardware > detection/configuration system. When we byte-compile these scripts, > they end up at about 400 to 500 k, which can add up.
Mmm. > The more important reason is the ease with which people can modify these > scripts, particularly when you are trying to fix a mis-configured > system. For people who are not really interested in programming, but > are just trying to get something to work, editing and running a script > is quite a bit easier than finding and downloading the source (which > maybe you can't do at all if your network isn't working) editing it, > building it, installing it. Granted, for most people, editing a script > is something they simply will not do. For a few others, building from > source is just fine. But there are some people whose input is important > to us who fall between these two categories. They have lots of energy > and intelligence, but they come from a world of GUI's and don't have the > Unix background we sometimes take for granted. If you point them to a > script and tell them "try some different values here and here" they will > go at it until it works. > > For myself, it is partly an aesthetic thing. Whenever practical, when > efficiency is not an issue, I like to distribute software in the least > obfuscated way, and put as few barriers as possible between the code and > the (hapless?) reader. Ok, you sort of convinced me. I would like the opinion of the other ocaml maintainers before i do that, in case there is something i may miss that they do not. That said, i could make a ocaml-interpreter package, which would be provided, replaced and conflicted by the ocaml package. What do you think of that ? Friendly, Sven Luther
