Mail Archives: djgpp-workers/1997/10/28/19:15:17
Some comments and concerns about these. I'm applying the patches, but
I'd like more work to be done on bringing these in line with the other
extensions.
On second thought, I'm backing them out. I can't accept half these
changes because they are have bad design, bad style and corrupt the
core of DJGPP's dos function set, plus other problems detailed below
(well, the problems I saw before I decided to take the patches back
out).
I'm willing to extend FSExt to include more core functions, but this
isn't the way to do it.
> 1997, Randall Maas
I can't accept something with a copyright that doesn't match djgpp's
copyright.
> # include <stdio.h>
Please use the same coding style as everyone else. No spaces after the #
> Content-Disposition: attachment; filename="ipc.h"
Please use the same template all the other includes use.
> diff -c src\libc\dos\io\_close.c _close.c
> *** \local\src\libc\dos\io\_close.c Sat Nov 25 18:47:06 1995
> --- \local\ginsu\diffs\_close.c Wed Oct 15 16:19:24 1997
> ***************
> *** 1,9
> ! /* Copyright (C) 1995 DJ Delorie, see COPYING.DJ for details */
> ! #include <unistd.h>
> ! #include <errno.h>
> ! #include <go32.h>
> ! #include <dpmi.h>
> ! #include <io.h>
> #include <sys/fsext.h>
>
> #include <libc/dosio.h>
>
> --- 1,8 -----
> ! /*
> ! emulates _close
> !
> ! 1997, Randall Maas rewrote this calls to use __FSEXT_Call
> ! */
> #include <sys/fsext.h>
> (more deleted)
Sorry, these are totally unacceptable. You may not remove or modify
my copyright notice. You may not replace the low-level DOS calls with
emulations. You must use the same coding style as everyone else.
> diff -c src\libc\dos\io\_close.txh _close.txh
Don't ever use backslashes when mailing me diffs. Patch wants to see
forward slashes.
> Content-Disposition: attachment; filename="_DOS_fsext.txh"
Avoid using names that don't fit in DOS's 8.3.
> int _nop(int handle)
This function name is too generic. After the next snapshot, please
submit diffs to rename it to something like _fsext_nop() instead.
> int _copy(const char* src, const char* dest)
Same here.
> ssize_t _ipc_pipe(int Transfer_FD, int filedes[2], int mode)
Same here.
> @node _link, file system
We *have* a function called link. If you want FSEXT to work right,
you have to hook that one so that programs will use FSEXT even when
they don't realize it. There's no benefit in writing a new API since
we'd have to go changing all the other programs too.
> off_t _lseek(int handle, off_t offset, int whence)
> void _pipe(int filedes[2])
> int _unlink(const char* src)
Same here.
> void __FSEXT_Emu_add(__FSEXT_Function* Func)
What's the point of this? We should only be extending functions that
already exist, not adding new ones that need to be emulated.
- Raw text -