Changes to the assembler headers

[Tony van der Hoff]

Philip Ludlam <philip at philipnet.com> wrote in message
<ad497d6b4c.philip at philipnet.com>

[snip]
> > ObjAsm has a 'feature' which seems to prevent it expanding path
> > variables from the command line.
> 
> I do hope Castle know about this 'feature' - and have removed it for the
> next release ;-) ?
> 
Not from me, they don't. I feel the path of least resistance is to try to
work 'round these niceties :-) However, careful reading of the ObjASM doc
suggests that this may be legitimate, i.e it talks about "-I directory", and
makes no mention of "path". 

The problem *really* shows itself up with the legacy "deep" directory
structure:

*show OSLib$p*
OSLib$Path(Macro) :
<OSLib$Dir>.,<OSLib$Dir>.Types.,<OSLib$Dir>.Macros.,<OSLib$Dir>.Core.,
<OSLib$Dir>.Computer.,<OSLib$Dir>.User.,<OSLib$Dir>.Toolbox.

*show OSLib$Dir
OSLib$Dir :
LanMan98::tony_lx.$.work.oslib-home.OSLib.!OsLib.release.deep.OSLib

The workround works even in this case :-)

[snip]
> 
> Now I'm trying BubbleHelp, and things aren't going to well either :-( .
> 
> I'm now getting: do objasm -desktop ^ -depend !Depend -throwback -nocache
> -I <OSLib$Path> -I <OS$Path> Gadgets.s.WriteBump o.WriteBump ARM AOF Macro

I've not seen this before. But I don't think you're allowed two -I's. I
think the second'll just overwrite the first. You could prove that by
swapping the two over.

There are two things you may try:

First, -I <OSLib$Path>,<OS$Path> should work. 

Secondly, set My$Path <OSLib$Path>,<OS$Path> and -I <My$Path>

Same thing, really.

[snip]
> 
> As I'm compiling Director OK I can be sure that there's no problem with -I
> <OSLib$Path> . And BubbleHelp did compile OK with OSlib 6.40. So to me it
> it looks like ObjAsm has another 'feature' which casues it to really fsck
> up filename references because even though the work directory is ^ (==
> Gadgets.s.^) it should be searching OSLib: for oslib.Hdr.OS, which it
> fails to do.
> 
> Am I right? If so, I'll send another bug report Castle's way :-( .
>
I can but agree with your interpretation, but then I can only see what
you've reported. I certainly agree that the ObjASM command line should be a
lot more flexible. Whether it constitutes a bug, well I just don't know.

Cheers, Tony

-- 
Tony van der Hoff
Buckinghamshire England