Hi, Roland Mainz wrote: >> This shouldn't be too surprising... >> > > Right... but the idea of pre-compiling the shell scripts is to reduce > the total CPU time consumed by the shell interpreter. Most scripting > languages (like Python) already do this by default. > > IMHO that does only make sens if the interpreter does this automatically.
For example in OS/2 REXX the tokenized (= more or less "compiled") version of a REXX script was saved in the extended attributes (EA) of the REXX script. Every time the interpreter executes the script it first checked for a tokenized version in the EAs. If there were no tokenized version the tokenized version was created by the interpreter, saved in the EAs and then executed. If there were already a tokenized version in the EAs the interpreter compared them to the source of the script. If the tokenized version was newer than the source code it was executed. If not, the REXX interpreter recreated the tokenized version, saved it in the EAs, and then executed the tokenized version. So, using this approach you have the advantages of "compiling" a script without using the advantages of an interpreter (e.g. I do NOT have to compile/link the script after every change in the source code) regards Bernd > ---- > > Bye, > Roland > > -- Bernd Schemmer, Frankfurt am Main, Germany http://home.arcor.de/bnsmb/index.html M s temprano que tarde el mundo cambiar . Fidel Castro