31–40 of 46 posts
Earlier quoted context omitted.
http://frama-c.com/
I think what flatestcat wants is something that actually functions and is proven to function correctly. Frama-C is yet another open source project that is long on hype but woefully short on delivery of actual, working code.
Can somebody recommend a book that covers recent techniques implemented in LLVM?
The good intro to compilers is Engineering a Compiler. It focuse more on optimization then most intro books, and its all SSA. If you want to go all in on SSA optimization, there is a book called Static Single Assignment Book and its written by a hole list of compiler writers. Its not finished but there is still a lot of information. You can find it here: http://ssabook.gforge.inria.fr/latest/book.pdf Or you can go wi…
[1] http://wingolog.org/archives/2011/07/12/static-single-assign...
[2] http://wingolog.org/archives/2014/01/12/a-continuation-passi...
why use c? its a horrific, logically incongruent lang
I won't echo this sentiment, but; why use C for writing a compiler ? It seems that languages in the ML family, among others, are nicer for that kind of domain. I guess it at least has its uses if you want your compiler to be really fast.
On the other hand, my projects in this vein are C++ projects but they implement functional languages or Scheme. I have implemented one parser that creates a syntax tree from a simple equational language, and several interpretive backends, such as SKI-combinator-based graph reduction and a TIM-based interpreter. I do care about the speed of the compilers I use for production code, but not for this work. The Scheme compiler I wrote was written to run on my own Scheme interpreter which was written in C++.
The reason I think most hobby compilers I know of don't go the way I've gone is that it takes tons and tons of research to find out how functional languages are implemented, and the Appel books are a bit daunting, too.
This book appears to be more of a "compiler-compiler design in C"; it goes through how to write a lexer and parser generator, then writes a compiler using them, and I think the resulting compiler is a bit of a letdown: it does not much more than translate C into a linearised subset of C, and so the IMHO more "interesting" and important parts of the back-end like register allocation and instruction selection are compl…
It's important to view this book in the context of its time. At that time, there were no books that showed you both the theory and complete running code for even this much. The compiler books (Dragon, 1st ed.; etc.) showed toy snippets but not a full lexer, parser, and code generator. There were only the articles in Dr. Dobb's about Small-C to show the way. So, at the time, Holub's book was a godsend. It married theory and implementation in a way that simply had not been done before. Hanson's and Fraser's 1995 book on a retargetable C compiler was a similar milestone, although it came out several years later.
After those two landmark books, explanations of full compiler implementations were no longer a rarity.
why use c? its a horrific, logically incongruent lang
I know what you mean, man. Lots of n00bz, ricers, and fanboiz will claim C is the fastest, most efficent lang because you can get "close to the metal" and "tweak". In reality, any code beyond trivial complexity will benefit much more greatly from algebraic rectification, which can only be done with certain languages that are amenable to formal analysis.
What exactly is "algebraic rectification"?
While it is generally true that having a formal semantics aids greatly in analysis, it is worth noting that a very large amount of program analysis work is targetted towards C. (And mind you, flexible, high level languages bring with it their own troubles. Analysis in the presence of higher order functions is not a panacea at all)
Can somebody recommend a book that covers recent techniques implemented in LLVM?
"There are advantages and disadvantages to an intermediate-language approach to compiler writing[...] Intermediate languages give you flexibility as well. A single lexical-analyzer/parser front end can be used to generate code for several different machines by providing different back ends that translate a common intermediate language to a machine-specific assembly language. Conversely, you can write several front ends that parse several different high-level languages, but which all output the same intermediate language. This way, compilers for several languages can share a single optimizer and back end."
Earlier quoted context omitted.
Until I see an award-winning 4k/64k demo written in one of these ultra-high-level languages, I stand by my opinion that C is more efficient.
Check your facts. A large majority of award-winning 4k/64k are written in C++.