News for hackers. Stuff that matters.
Re:The Good, the Bad, and the Ugly: The Unix Legacy (by Rob ... by silentbicycle on Wednesday May 06, 2009 @08:50PM (#597161)
Elsewhere in the plan9 writings, Rob Pike says that he structured it by putting all of the stuff that would need to be ported in one group of files (which export the same interface), conditionally compiling the relevant subset of those, and then writing everything else in terms of those. (A better packaging/namespace system could handle this much more elegantly, though.)

Also, the C preprocessor is largely ignorant of C syntax. Am I the only one who finds that incredibly strange?

Re:The Good, the Bad, and the Ugly: The Unix Legacy (by Rob ... by philwelch on Thursday May 07, 2009 @03:16AM (#597588)
If the preprocessor knew the syntax, it would happen after parsing rather than before, which is exactly how every language other than C handles things that C handles using the preprocessor (turning "using" or "include" or "import" into keywords instead of preprocessor directives). I'm really not sure in what way it would be a pre-processor at that point.

So perhaps it's more accurate to say preprocessors aren't particularly useful.

Re:The Good, the Bad, and the Ugly: The Unix Legacy (by Rob ... by vorador on Thursday May 07, 2009 @12:29AM (#597415)
What would be the advantages of a preprocessor knowing the syntax ?
Re:The Good, the Bad, and the Ugly: The Unix Legacy (by Rob ... by silentbicycle on Thursday May 07, 2009 @09:35AM (#597885)
While it probably wouldn't be as useful as Lisp macros, due to the comparatively complex syntax, it would at least avoid bugs involving things like #define macros that evaluate their arguments multiple times.

OTOH, not having the C pp tied to C means it can be used for other stuff. The only thing I can think of offhand that uses it is the BSD calendar system, though, and that's largely just #include.