blob: 217413346b1797b09f15be8787093a7b16e35df1 [file] [log] [blame]
index of the source files:
README This file.
LT.diff A patch for libtool.m4
ltstr.def The `glue' definition file that sources the following:
lt_base.def Global variables used for all operation modes
lt_clean.def `clean' mode script
lt_compile.def `compile' mode script
lt_execute.def `execute' mode script
lt_finish.def `finish' mode script
lt_install.def `install' mode script
lt_link.def `link' mode script
ltopts.def The definitions for the ltmain options
ltmain.c The main routine for the new ltmain (aka libtool).
ltmain.mk The make file for all this
ltmain.tpl The template for the shell script (ltmain.in) variation
ltstr.tpl The template for producing the compiled versions of
the scriptlets defined in the lt_*.def files.
================
These are the generated files:
ltconfig.x Generated for a particular configuration via the
LT.diff patch, above.
ltmain.in Generated version of the traditional file
ltopts.c Generated option definition structure
ltopts.h Option interface file
ltstr.c The strings for emitting the shell commands
ltstr.h The interface file for same
================
The executable requires -lopts to be able to link.
The latest autogen archive ought to have a K&R friendly version
of the option processing library that can be compiled and linked
into the executable. Of course, AutoGen-5 is required to be able
to create the last five generated files, so you should
have it available. :-)
================
I have not done any timing tests, but at this point any savings
ought to be quite marginal. About all that is saved is interpretation
time, very few forks and execs. That all comes later. :-)