> C-files in the OCaml-distribution are usually compiled with -fno-defer-pop
> when they are supposed to be used with byte code, too. What is the
> rationale behind?

There used to be a bug in gcc whereas code using computed gotos (such
as the OCaml bytecode interpreter) would be incorrectly compiled
unless the -fno-defer-pop option is given.  I don't know if this has
been fixed in recent versions of gcc.

At any rate, you're correct that -fno-defer-pop is needed only for
compiling byterun/interp.c, and applying it to all bytecode-related C
files is a bit excessive.

> Is this only intended to limit stack growth a bit or
> are there other reasons? This opposite of this option is normally turned
> on when optimisation (-O) is used. Would you recommend this setting
> generally for C-code that should cooperate with OCaml-byte code?

No, it shouldn't be necessary.

- Xavier Leroy
-------------------
Bug reports: http://caml.inria.fr/bin/caml-bugs  FAQ: http://caml.inria.fr/FAQ/
To unsubscribe, mail caml-list-request@inria.fr  Archives: http://caml.inria.fr