On 12 Aug 2009, at 6:09am, Roger Binns wrote: > If you read Bernstein's retrospective on qmail, one of his > suggestions is to > write code in a higher level language (especially more secure in the > sense > that C isn't such as preventing buffer & integer overflows etc) that > is then > transcoded to C. (Some conspiracy theorists looking at his code > claimed he > actually did do this :-) > > So in theory given unlimited CPU and memory it should be possible to > take > the SQLite C code and turn it into something higher level and then > turn that > back into the more verbose platforms (.NET, Java). The big > advantage of > this approach (after the upfront work) is that keeping up with SQLite > progress is easy.
The job of translating from one computer language to another hasn't been solved yet. You'd have thought that it would, wouldn't you ? After all, it's a pure computing problem which requires only geeks who know a couple of computer languages well. But it hasn't, partly because good programmers adopt styles which suit the language they're writing in and it turns out that styles do not translate well into other languages. .NET and Java and C# and C++ and Objective-C are all C-like languages. They have many of the faults that C does and lack the portability. I suspect that Python is the most popular language which is high enough above C to do this properly. You're write a Python2C translator, then use a compiler that had extremely good optimisation routines and hope that it gave you reasonably small-and-fast resulting code. We no longer have the heritage of ubiquitous high-level languages. Fortran, Pascal, Structured Basic, and everything like that have gone. We're looking to Python and Ruby and similar languages to fill the gap but they're lower than we're used to. Simon. _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
