From: Bruce B. <ba...@en...> - 2004-03-26 20:45:34
|
MJ Ray wrote: > > A lot of cookbook stuff need not use PLT-specifics. The pregexp and SRFI > stuff will probably work just as well elsewhere. I think marking > PLT-specific to allow other versions later is a very good idea. Of > course, we should not shy away from including useful PLT-specific parts. > I agree that there should be a "Scheme" cookbook as well as a "PLT-Scheme" cookbook. Knowing that there aren't infinite resources avaliable for writing these I would prefer emphasis on the latter since I can find a fair amount of non-PLT scheme code around the web. That is not to say that the PLT cookbook should use gratuitous "PLT-only" idioms when an equivalent exists in R5RS, but I'm not aware of many of those (square-brackets notwithstanding). But as a working developer I would rather see something like case-lambda to express variable argument functions than the (very roughly equivalent) improper list method. Same goes for modules, units, structures as procedures, etc. > Bane of my life is trying to make Scheme code portable. We may all have > PLT on our systems, but sometimes you have to run under something else. > The pigheaded refusal of some good developers to support that is still a > drawback of Scheme. > No, at least on my part it isn't pigheadedness, it's the decision I made to use PLT for doing large system development. The argument has been made elsewhere that R5RS defines not a language but a template that can be expressed differently in implementation. For instance I also use Sisc since I support a large java servlet project and it is easier to integrate. I certainly use the knowledge I have acquired from working with PLT there but I have to think about structuring my code differently because of it's (more limited in my view) module system. Turns out there is a Sisc cookbook on their wiki and it has a few nice recipes on it. What a great idea! -- For every expert, there is an equal but opposite expert -- Arthur Clarke |