ids. This way, we can delay loading spans until we know that we definitely
need an object file.
Did some restructuring for writing of span lists to the debug info file.
git-svn-id: svn://svn.cc65.org/cc65/trunk@5261 b7a2c559-68d2-44c3-8de9-860c34a00d81
read when the -t or -C switch is encountered and parts of it are processed.
The remaining parts are processed when all object files and libraries have
been read. To make this work, the expression evaluation in cfgexpr has been
rewritten to generate true expression trees. This means that expressions in
the linker config may use exports from the object files.
Separation of config file processing is the base for several enhancements, for
example forced imports by linker config.
This code needs more work and is only very, very, very roughly tested.
git-svn-id: svn://svn.cc65.org/cc65/trunk@4840 b7a2c559-68d2-44c3-8de9-860c34a00d81
Use the address size for import, export and debug symbols (object code
change).
More changes to support the --memory-model switch and address sizes.
git-svn-id: svn://svn.cc65.org/cc65/trunk@2691 b7a2c559-68d2-44c3-8de9-860c34a00d81
string table in the object file that (currently) holds all identifiers
from the import, export and debug info sections. The plan is to put all
strings into this table, so we have them in a central place and don't
waste memory. Apart from that, the indices are unique, so comparing strings
should be a lot easier than before (as soon as the programs take advantage
of this fact, which is currently not the case).
git-svn-id: svn://svn.cc65.org/cc65/trunk@2169 b7a2c559-68d2-44c3-8de9-860c34a00d81
all sorts of things in the object files. This does not only make the object
files smaller, but does also remove several limits (strings may be longer
than 255 bytes, several counters no longer have 8 or 16 bit limits).
git-svn-id: svn://svn.cc65.org/cc65/trunk@260 b7a2c559-68d2-44c3-8de9-860c34a00d81
