Newsgroups: comp.sys.amiga.programmer Path: utzoo!utgpu!bhgs From: bhgs@gpu.utcs.utoronto.ca (R. Holdom) Subject: Re: ARP and DICE - How to link?? Message-ID: <1991Mar6.213253.27365@gpu.utcs.utoronto.ca> Organization: UTCS Public Access References: <1991Feb25.121623.16551@uni-paderborn.de> Date: Wed, 6 Mar 1991 21:32:53 GMT In article estdwha@warwick.ac.uk (CrisP) writes: (stuff deleted) > >With Dice2.06 there is a little program call fdtolib that can create the such >a .lib file from the .fd file. I have already created myself a amigs.lib from >.fd files. And am working an arp.lib. But I have forseen a problem with the >PrintF() and simular functions. In the Arp library you pass the mutiple >undefined parameters as a pointer to an array. It would be nice to pass them >as seperate parameters. I know that fdtolib creates the glue code needed from >the .fd file. I know that it then uses das to assemble it. But what I can't >work out is what it does from there. > >~ CrisP. (more stuff deleted) Just this mourning I was doing this very thing. fdtolib works fine for all of the simple function calls (ie no Printf's and stuff that don't return a second value via IoErr()) For the rest, the arp guys have already done the job for us. In the programers package (the one with the documentation) there is a file called LatticeGlue.a which was designed to work with Lattice (as you may have guessed) but which suits our needs fine. The only thing I had to modify was they have the line: xref _ArpBase but this causes the assembler (at least the one I used) to make this symbol by refferenced from the pc register, instead of a4, as we want. So i just put the line: section "",data right before the above line and everything was fine. I also took out the stuff for making the choice between Near and Far memory models, since we want the near one. I assembled the file using the Lattice assembler from Lattice C 4.0.1, since Matt's is still somewhat anemic(sp?). Then I just joined the two files, and everything is fine. If anyone is interested, it probably wouldn't be too hard to make a file that Matt's assembler could handle. So send me a note and I'll try it. Jon Spencer.