Re: srfi-0 and srfi-7
"Neil W. Van Dyke" <[email protected]>
| Newsgroups | gmane.lisp.scheme.plt.schematics |
|---|---|
| Message-ID | <[email protected]> |
> SRFI 7 is for writing portable programs based on specifications shared > by multiple implementations. Adding "MzScheme-specific" extensions > defeats that goal, and SRFI 7 just isn't good at handling this case. I think "defeats" might depend on whether you are talking technically or politically. Technically, it's just a namespace and syntactic conditionals. > That of, course, is not only unportable now, it's also likely to make > a program written and working now fail in the future. I think I appreciate your point. However, libraries, if useful, would be actively maintained, and therefore could track the changing universe of implementations. There is just not enough person-power available to write excellent SRFIs and implementations of same right now. That reality cannot be forced politically, because there is little leverage, and few agents to influence anyway. Also, in practice, portability abstractions that are expressive and evolutionarily responsive enough to capture all the pertinent information on which crucial implementation/portability decisions might be made in programs-- are often impractical. You start bumping into intractable problems of ontology and decidability, and also get maintenance nightmares of the type you were trying to avoid. I think there should also be no reason, in a language with Scheme's syntactic power, that I can't have a single source code file that captures decisions I have made about how to get it to work well with different extensions of several Scheme implementations. If we want a commoditized, suffocating language, there is Python. :) I think I can make my code much more portable and widely useful by judicious reference to non-SRFI "features" than I can by imagining standards and libraries that do not exist and that nobody has time to write. However, if I, and many other modest-contributors like myself, eventually hit a critical mass of library usefulness, then maybe sufficiently good portability abstraction libraries we need will get written. Right now, the Scheme universe consists of a few-dozen poorly-funded research projects, pedagogical religions, and spare-time hobbies, however, and SRFIs and their implementations are poor for much of what they are supposed to be. ------------------------------------------------------- This SF.Net email is sponsored by the new InstallShield X. From Windows to Linux, servers to mobile, InstallShield X is the one installation-authoring solution that does it all. Learn more and evaluate today! http://www.installshield.com/Dev2Dev/0504