On August 21, 2003 05:08 pm, Martin Wilck wrote:
You challenged me :-) really, shell coding is one of my favorites.
This is very nice Martin, I'll grant you that. However I'm afraid I'm not sold. And I'm saying this without any emotional attachment to my code :). But let me explain why: -- I think we're ending up raising the bar rather than lowering it. For most developers, winegcc.c is a tiny C program they can easily understand and modify. For the same group, yours is a non-trivial script. -- winegcc.c is portable and will run without problems on all systems, even on Windows. On the other hand, for scripts of this size it's very easy to add features that will break on some implementations, or that are bash specific. Moreover, we don't find out about them until a user tries to run the program, at runtime. -- the result is not smaller. Some things are cleaner, true, but I don't think the .c program was that unreadable :) -- last, but not least, the thing is for sure slower. Maybe not by much, but I think it's measurable, once we compile Wine with winegcc.
I, for one, am more confortable with the C semantics, rather than sh. I can right sh scripts, but I'm not in my element doing non trivial stuff (or maybe I'm more confortable in C :)).
Anyway, just my .02$ -- I guess it's Alexandre's call.