The experiments with "publishing" new API routines in the ui library (#1059) that have been written in GOC (probably for the first time in GEOS history...) indicate that the published routines are not correctly extracted from the object code: disassembly shows that the resulting segment in the target geode is filled with zeroes.
If publishing works correctly, a copy of the original routine should be placed into a separate segment in the geode.
This is probably an incompatibility between glue and the object format (or debug info) generated by Watcom.
The experiments with "publishing" new API routines in the
uilibrary (#1059) that have been written in GOC (probably for the first time in GEOS history...) indicate that the published routines are not correctly extracted from the object code: disassembly shows that the resulting segment in the target geode is filled with zeroes.If publishing works correctly, a copy of the original routine should be placed into a separate segment in the geode.
This is probably an incompatibility between
glueand the object format (or debug info) generated by Watcom.