delorie.com/archives/browse.cgi   search  
Mail Archives: djgpp/1997/08/13/13:04:09

From: Lyle <lpak1 AT NO_SPAMccds DOT cc DOT monash DOT edu>
Newsgroups: comp.os.msdos.djgpp
Subject: Re: Perculiar bug fix?
Date: Wed, 13 Aug 1997 20:46:19 +1000
Organization: Monash Uni
Lines: 61
Message-ID: <33F1907B.E4196932@NO_SPAMccds.cc.monash.edu>
References: <33EEE76F DOT 2F4B904 AT NO_SPAMccds DOT cc DOT monash DOT edu> <01bca6b4$ae002360$45111d0f AT p1675cra DOT rose DOT hp DOT com>
NNTP-Posting-Host: ascend-1-30.cc.monash.edu.au
Mime-Version: 1.0
To: djgpp AT delorie DOT com
DJ-Gateway: from newsgroup comp.os.msdos.djgpp

Hi,

Andrew Crabtree wrote:
> 
> > the exact same code, compiled under Borland TC++ V3.0? Is this normal?
> > Is there something i;m doing wrong (most certainly!).
> Were you accessing data via structure pointers??  If so note that djgpp
> tends to pad data more than some of the older dos compilers,
> and the default size of 'int' is 32 bits, not 16.  You could try using
> __attribute__ ((packed)) to get rid of the padding, and replace ints with
> shorts as needed.  See the docs on gcc extensions for packing data.  It may
> also be in the FAQ.

ahh i see - that would explain why the structures were being read weird.
Not a problem, i just deleted the files and started again. I surmised
something was different, but i'm thankfull to now know what :)

> 
> > I still also have the problem that i cannot debug any of my code that i
> > have '#include' into modules? Do i have to link all my source files,
> > instead of just all my modules?
> 
> Putting code in header files is generally a poor idea.  I suspsect you have
> run into one of the limitations of sdb debugging format.

The files are not header files, but source code. i.e, i have an object
with approx 5,000 lines of code. To mkae it easier for me to debug, i
simply split the object further up into more source files. I'm sure
there is nothing wrong with that!

> The easist way to do what you want (shameless plug following) is to get
> pgcc from www.goof.com.  Its fairly stable right now, with a few quirks
> with known workarounds.  Just compile your code with -gstabs, and as long
> as you use gdb you should be OK.  Another option would be
> to get Robert's patch to gcc and recompile it yourself.
> 
> Andrew

Robert's Patch? I'm using RHIDE (lattest ver as i have checked
recently)? If i put gstabs on should it work?

I just tried it with -gstabs+, i got the error "-gstabs+ not supported
by this GCC". So i tried it with -g ang -gstabs+ and it worked. However,
i still could not trace into my sub-source code. The only thing i can
assume is that the -g took precedence and, ineffect, the -gstabs+ did
nothing??

Thanks for your contructive reply!

Cheers,
	Lyle

-- 
NOTE: Remove The comment "NO_SPAM" To Reply via Email!

-------------------------------[ **NEW ADDRESS**
lpak1 AT ccds DOT cc DOT monash DOT edu DOT au]
" Hello Chevra Kadish, You Kill 'em, We Chill 'em "            .----,
						               | oO |
	HTTP://www-personal.monash.edu.au/~lpak1/              | \/ |
		                                               `----'

- Raw text -


  webmaster     delorie software   privacy  
  Copyright © 2019   by DJ Delorie     Updated Jul 2019