While listening to Paul Edwards Pete heard:tobruk live stats



G'day Paul



 PC> 1st one is in module 'faddr.c' and is 'dir.h' does not exist

 PC> however



 PE> You probably mena trav.c, no?



Yes your correct it was in trav.c, faddr.c has the 'segment lost in

conversion' error.



 PC> this just seems to be a wrapper for direct.h anyway so can be

 PC> overcome by me just creating a dir.h that just has '#include

 PC> ' in it or defining _QC25_ and then #ifdef _QC25_ , i

 PC> think the first option is the better one.



 PE> There is no directory traversal for Quick C.  The default is to

 PE> compile with the ability to only process a single .PKT file.



the QC direct.h has chdir() getcwd() mkdir() rmdir() profiled in it so

i assume these would be directory traversal functions. The dir.h

header file in the Symantec compiler seems to be just a wrapper for

direct.h below is the Symantec dir.h 



#ifndef __DIR_H

#define __DIR_H



#include 



#endif



 PE> However, you must either be defining MSDOS yourself, or it is

 PE> being defined automatically.  Try adding "-UMSDOS" to the

 PE> command



MSDOS is being defined automatically, i'll try undefining it and see

how it goes.



 PE> line when compiling trav.c.  Or you should be able to do a

 PE> -UMSDOS on all files actually.



Yep just tried -UMSDOS and it removed the above errors, now there is

only the 'segment lost' errors to contend with.



 PC> 2nd one is warnings about 'segment lost in conversion' on lines

 PC> with things like '(void)faddr;', '(void)mailproc;' and

 PC> 'unused(repo);' on them although i'm compiling in the large

 PC> memory model.



 PE> Ok, you'll have to live with the warnings there, or else change

 PE> the unused() macro to not do anything.  I'll see how much effort

 PE> is involved in changing all the others to call unused().



 PC> 3rd one is in trav.c and is " 'ffblk' uses undefined

 PC> struct/union 'ffblk' " on line with 'struct ffblk ffblk;' on it,

 PC> a few other errors because of this undefined struct/union.



 PE> That's the same problem with -UMSDOS when compiling trav.



 PC> 4th one is also in trav.c about findfirst() findnext() 'no

 PC> function prototype given'.



 PE> Ditto.



 PC> The good news is Tobruk compiled without errors using Symantec

 PC> C++ 6.1 so looks like i'm stuck using a windows hosted IDE :-(

 PC> next step is to throw a few *.pkt files at it and see how it

 PC> goes.



 PE> Thanks for the feedback.



 PC> BTW why do you think QC choked on the code, old age maybe, has

 PC> the C std changed that much.



 PE> findfirst() is non-standard, which is why by default it is not

 PE> compiled in.



There are file finding functions in QC _dos_findfirst()

_dos_findnext() i'd assume the naming conventions would carry across

all the MS compilers so maybe a macro would work, the MS syntax is



unsigned _dos_findfirst( char *filename, unsigned attrib,

                         struct find_t *fileinfo );



unsigned _dos_findnext( struct find_t *fileinfo );



something like



#define findfirst(a,b,c)  _dos_findfirst(a, c, b)





   -=Pete=-





--- Blue Wave/Max v2.12