System image: Difference between revisions

Revision as of 06:48, 3 April 2018

Revision as of 21:00, 17 July 2018

Line 16:

===Programming language support===

Some [[programming language]]s provide a command to take a system image of a program. This is normally a standard feature in [[Smalltalk]] (inspired by [http://www.mprove.de/diplom/gui/kay68.html FLEX]) and [[Lisp (programming language)|Lisp]], among other languages. Development in these languages is often quite different from many other programming languages. For example, in Lisp the programmer may load packages or other code into a running Lisp [[implementation]] using the [[read-eval-print loop]], which usually compiles the programs. Data is loaded into the running Lisp system. The programmer may then [[ROM image|dump]] a system image, containing that pre-compiled and possibly customized code - andcode—and also all loaded application data.

<!--

It's to avoid the time penalty of recompiling that software like Emacs use images, and modern Common Lisp interpreters, anyway, compile packages as a matter of course (See SBCL, CMUCL, etc.)

-->

Often this image is an executable, and can be run on other machines. This system image can be the form in which executable programs are distributed — thisdistributed—this method has often been used by programs (such as [[TeX]] and [[Emacs]]) largely implemented in Lisp, Smalltalk, or [[idiosyncratic]] languages to avoid spending time repeating the same initialization work every time they start up.

Similar, [[Lisp Machine]]s were booted from Lisp images, called Worlds. The World contains the complete operating system, its applications and its data in a single file. It was also possible to save incremental Worlds, that contain only the changes from some base World. Before saving the World, the Lisp Machine operating system could optimize the contents of memory (better memory layout, compacting data structures, sorting data, ...).