Message-ID: <3f51fb97$0$164$cc7c7865@news.luth.se> From: Martin Str|mberg Subject: Re: as and structs Newsgroups: comp.os.msdos.djgpp References: <3f3e2079$0$168$cc7c7865 AT news DOT luth DOT se> <3f3e4fcf$0$172$cc7c7865 AT news DOT luth DOT se> User-Agent: tin/1.4.6-20020816 ("Aerials") (UNIX) (NetBSD/1.6Q (alpha)) Date: 31 Aug 2003 13:43:51 GMT Lines: 36 NNTP-Posting-Host: speedy.ludd.luth.se X-Trace: 1062337431 news.luth.se 164 130.240.16.13 To: djgpp AT delorie DOT com DJ-Gateway: from newsgroup comp.os.msdos.djgpp Reply-To: djgpp AT delorie DOT com Hans-Bernhard Broeker wrote: : Martin Str|mberg wrote: :> source then? E. g. DEBUG in my example. :> :> 1. What should I do to make this work nicely with prototypes and :> :> structs in header files? :> : Enclose the C-only parts in some #ifdef, e.g. #ifdef __GNUC__ or :> : whatever. :> Which define makes a good C<->assembly discriminator? __GNUC__? : I should think so. 'gcc -v' on both a .c and a .S file should : enlighten this further. Thanks. This is what I did. Case 1, some code for DJGPP libc. A sed script that extracted the relevant #defines into a local header file (thank you, Richard Dawe) and use the created header file in the .S files. Case 2, my own pet project. Surround the structs with #ifdef __GNUC__ and #endif (there's only one 32bit header file with this and it's automatically generated as a glue file between 16bit djasm code and 32bit code). : That expectation would be quite wrong, though. The choice of flags : depends on the source language (via the file extension), not the : program being called to translate them. Ok. Thanks! Right, MartinS