Cook 2.25

Cook is a tool for constructing files. It is given a set of files to create, and recipes of how to create them. In any non-trivial program there will be prerequisites to performing the actions necessary to creating any file, such as include files. Cook provides a mechanism to define these.

When a program is being developed or maintained, the programmer will typically change one file of several which comprise the program. Cook examines the last-modified times of the files to see when the prerequisites of a file have changed, implying that the file needs to be recreated as it is logically out of date.

Cook also provides a facility for implicit recipes, allowing users to specify how to form a file with a given suffix from a file with a different suffix. For example, to create filename.o from filename.c

Cook is a replacement for the traditional make(1) tool.
There is a make2cook utility included in the distribution to help convert makefiles into cookbooks.

Cook is more powerful than the traditional make tool.

Cook has true variables, not simple macros.
Cook has a simple but powerful string-based description language with many built-in functions. This allows sophisticated filename specification and manipulation without loss of readability or performance.

Cook has user defined functions.

Cook can build in parallel.
Cook is able to build your project with multiple parallel threads, with support for rules which must be single threaded. It is possible to distribute parallel builds over your LAN, allowing you to turn your network into a virtual parallel build engine.

Cook can distribute builds across your LAN.

Cook is able to use fingerprints to supplement file modification times. This allows build optimization without contorted rules.
Cook can be configured with an explicit list of primary source files. This allow the dependency graph to be constructed faster by not going down dead ends, and also allows better error messages when the graph can't be constructed. This requires an accurate source file manifest.

In addition to walking the dependency graph, Cook can turn the input rules into a shell script, or a web page.

Cook runs on almost any flavor of UNIX. The source distribution is self configuring using a GNU Autoconf generated configure script.
Cook has special cascade dependencies, allowing powerful include dependency specification, amongst other things.

See the README file for a description of the new features and bug-fixes in this release. This is also in the Cook Reference Manual, along with the build and installation instructions, and all of the manual pages.

If you are putting together a source-code distribution and planning to write a makefile, consider writing a cookbook instead. Although Cook takes a day or two to learn, it is much more powerful and a bit more intuitive than the traditional make(1) tool. And Cook doesn't interpret tab differently to 8 space characters!

Cook is written and owned by Peter Miller and is freely distributable under the terms and conditions of the GNU GPL.

Cook is developed using Aegis, a transaction based software configuration management package.

There is more Software by Peter Miller at his home page.

This page has been accessed approximately

times since 17-Jul-2004.

Cook is a replacement for the traditional make(1) tool.	There is a make2cook utility included in the distribution to help convert makefiles into cookbooks.
Cook is more powerful than the traditional make tool.
Cook has true variables, not simple macros.	Cook has a simple but powerful string-based description language with many built-in functions. This allows sophisticated filename specification and manipulation without loss of readability or performance.
Cook has user defined functions.
Cook can build in parallel.	Cook is able to build your project with multiple parallel threads, with support for rules which must be single threaded. It is possible to distribute parallel builds over your LAN, allowing you to turn your network into a virtual parallel build engine.
Cook can distribute builds across your LAN.
Cook is able to use fingerprints to supplement file modification times. This allows build optimization without contorted rules.	Cook can be configured with an explicit list of primary source files. This allow the dependency graph to be constructed faster by not going down dead ends, and also allows better error messages when the graph can't be constructed. This requires an accurate source file manifest.
In addition to walking the dependency graph, Cook can turn the input rules into a shell script, or a web page.
Cook runs on almost any flavor of UNIX. The source distribution is self configuring using a GNU Autoconf generated configure script.	Cook has special cascade dependencies, allowing powerful include dependency specification, amongst other things.