You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Is your feature request related to a problem? Please describe.
Currently, the struct definitions and the typedefs that go with them are generated into a single header file.
This can cause a backward-compatibility problem, in particular for the "msg" module, which defines an abstract base typedef for its messages here:
As the generated EDS file also has a typedef in it, this results in a duplicate/conflicting type definition. (not for the struct, but just the typedef).
Describe the solution you'd like
This could be solved if the EDS-generated defs were in separate files, such that the structs could be defined without the typedefs, or vice versa.
Describe alternatives you've considered
The underlying issue is that this currently requires a bit of a hack to get the EDS-generated definitions to take the place of the traditional definitions. Goal is to remove the hack, so any solution that gets there is worth considering.
Additional context
Ideally the EDS-generated headers should conform to the existing CFE headers, minimizing/avoiding any change on the CFE side where possible to do so.
Requester Info
Joseph Hickey, Vantage Systems, Inc.
The text was updated successfully, but these errors were encountered:
jphickey
added a commit
to jphickey/EdsLib
that referenced
this issue
Feb 29, 2024
Implements a bit of paradigm shift in how names of symbols/types are
generated, relying on prefixes rather than suffixes. Using suffixes
has an opportunity of name collisions that is not possible when using
prefixes.
A separate typedef file maps to the existing type names and thus keeps
it compatible with existing code, but only if that typedef file is used.
Implements a bit of paradigm shift in how names of symbols/types are
generated, relying on prefixes rather than suffixes. Using suffixes
has an opportunity of name collisions that is not possible when using
prefixes.
A separate typedef file maps to the existing type names and thus keeps
it compatible with existing code, but only if that typedef file is used.
Is your feature request related to a problem? Please describe.
Currently, the
struct
definitions and the typedefs that go with them are generated into a single header file.This can cause a backward-compatibility problem, in particular for the "msg" module, which defines an abstract base typedef for its messages here:
https://github.com/nasa/cFE/blob/8cdad66e4eafd11c2ea93d0360867045b77fa981/modules/core_api/fsw/inc/cfe_msg_api_typedefs.h#L107
As the generated EDS file also has a typedef in it, this results in a duplicate/conflicting type definition. (not for the struct, but just the typedef).
Describe the solution you'd like
This could be solved if the EDS-generated defs were in separate files, such that the structs could be defined without the typedefs, or vice versa.
Describe alternatives you've considered
The underlying issue is that this currently requires a bit of a hack to get the EDS-generated definitions to take the place of the traditional definitions. Goal is to remove the hack, so any solution that gets there is worth considering.
Additional context
Ideally the EDS-generated headers should conform to the existing CFE headers, minimizing/avoiding any change on the CFE side where possible to do so.
Requester Info
Joseph Hickey, Vantage Systems, Inc.
The text was updated successfully, but these errors were encountered: