On Thu, Jan 22, 2009 at 03:38:37PM -0500, Andrew Gwozdziewycz wrote: > Certainly true, but let me ask you this. If you were planning to use an ASCII > file as a storage engine, what details would you include? Would you not > include a way to join other ASCII files, or do some sort of search? Would > you not build in 'grep' to do filters with regex?
Probably. But I'd probably also store data in a hierarchical fashion if I were using S-expression data in files (ie, nested trees). If I knew I was using a RDBMS, I'd use a totally different data model, something that maps well to relational queries. > The benefit of abstraction here is that we don't care about where it's stored. > If we care about performance, we're going to use the engine that best fits > the features we need anyway. We're not going to use ASCII files if we set > our data up in an entirely relational way, This is my point; you just said yourself that you care (or are forced to care) about how it's stored. You can't abstract away from that. So switching to a different storage backend won't do you much good; you've already set things up in a non-relational way so the RDBMS won't be able to handle it very well. > but there's no reason to not simulate > a feature that doesn't exist in an storage engine if it makes sense to use. It makes sense up to a certain point. It might be worth an experiment to see how far you can get with this. You might even succeed and have a great and pluggable data storage backend. I'm certainly not going to stop you. Cheers, Peter -- http://sjamaan.ath.cx -- "The process of preparing programs for a digital computer is especially attractive, not only because it can be economically and scientifically rewarding, but also because it can be an aesthetic experience much like composing poetry or music." -- Donald Knuth
pgp3Wx5aZBVBt.pgp
Description: PGP signature
_______________________________________________ Chicken-users mailing list [email protected] http://lists.nongnu.org/mailman/listinfo/chicken-users
