AWeb doesn't compile

[Minuous]

Hello all,

I'm trying to compile AWeb 3.5.10, which was apparently only released as source code, with a view to at last doing a release of the AWeb 3.5.10 binary and then making further improvements and bug fixes, to create 3.5.11 and beyond.
  It's supposed to compile OK but it doesn't.
  SAS/C is giving lots of varargs- and pragma-related errors. How could this have been successfully compiled on the original developer's system but not on mine?

Also, it seems to be expecting a file "awebgst" which doesn't exist, does anyone have a copy of this or is there some way I can generate it myself?

I tried to join the AWeb eGroup on Yahoo but it requires approval and the moderators don't seem to ever go online anymore. Does anyone have an archive of the files that are there?

[matthey]

AWeb is a pain to compile for 68k. I never got it to compile completely. I recommend GCC 2.95.3 as that's what most of the developers were using and it generates pretty good code. Besides some varargs problems (I reported 1 at the time), it's important to have the correct include files as some of the 3rd party include files have bugs including the AmigaOS 3.9 includes. Some of the last authors like broadblues (AmigaOS 4 Blender porter on http://www.amigaworld.net) and Opi (MorphOS user) are still around and would probably help you. There's a couple of AWeb guys on http://www.natami.net/ too. There has been several threads and talk of restarting active development also. Sign up, post a new thread and I think you will find that you will have plenty of friendly help.

[lsmart]

If I am not mistaken, there is no point in using "extern" on a function declaration. So I guess you have to understand, what the Varargs macro does. Perhaps you can just remove the storage class specifier.
Have you looked at the code after preprocessing?

[Minuous]

Yeah, I've sorted out that particular issue now. It's just a symptom of the larger issue, which seems to be that all the SAS/C code has been removed and replaced with OS4 crap. So probably the best thing to do would be to start with the original 2002 source release, which *did* work with SAS/C, and fork it from there. :-(

[matthey]

I didn't expect it would compile any more with SAS/C but the last versions did compile with GCC 2.95.3 and mostly with later versions of GCC. Sure, the conditional code for 68k, AROS, AmigaOS 4.x and MorphOS is going to make the source harder to read and create some problems but there has been some significant *positive* changes since the 2002 release. Also, GCC 2.95.3 generates better code than SAS/C in my opinion.

[Minuous]

OK, I've now successfully compiled the original 2002 codebase. So that is a starting point for compiling 3.5.10.

I don't have a problem with the conditional code. My own programs are the same, they have conditional code to support OS3/OS4/MOS and where feasible AROS. What I'm annoyed about is the removal of the existing SAS/C code, it means that a job that would have been fairly straightforward is going to be a massive PITA. There's not even any StormC support, even though StormC can run in GCC mode :-(