This is the mail archive of the gcc-bugs@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: -fsquangle


> However, I do not understand how `squangle' works. Is there
> a runtime penalty associated with it? If not, why not change
> to `squangle' mode by default? 

There is no runtime penalty. It is just a different mangling
algorithm. There are different reasons why it is not default:

- last I heard, it was not complete (i.e. the spec defining it in
  gxxint.texi was not fully implemented). I'm not sure whether this
  is still the case.
- demangling it is incomplete/broken
- it is not binary compatible

> If binary compatibility is an issue, make it an option in the
> configure script and advertise it in the INSTALL docs.

This may happen with future releases. However, there are also pending
changes that break binary compatibility and which are not detected by
the linker. This will be a real pain, since people won't recompile
everything even if they have been warned with blinking letters.

If these changes are released together with squangling, people will
more get linker errors if they mix binary-incompatible objects.

Regards,
Martin


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]