Mail Archives: djgpp/1995/05/17/05:36:25
> Here are the executables contained in the following
> utility files on oak.oakland.edu /pub/simtel/msdos/gnuish
> [thanks to Aaron Ucko for pointing me at tut111ax]:
>
> fut312bx.zip (file utils)
Just FYI:
Whoever uses this port of GNU Fileutils, should excercise care when
working with networked (NFS or Novell) drives. The executables use
a custom version of stat() which gets the number of the first cluster
of a file as a substitute for the inode number which many of these
utilities use to decide if any two files are actually the same file.
This is fine, but that particular version of stat() doesn't have
backup solution for networked drives and for empty files on local
drives, where the cluster is unavailable. So if you ever see a
message telling you that two different files are the same (e.g., cp
will refuse to copy a file into itself), you will know...
Uh-oh. I don't do this stuff under DOS anymore, so it's been a long
time, but I get a bad feeling. Has anybody tried to use this package
to compile itself using the FSF makefiles? In particular, I remember
when I tried compiling some of the utilities myself using Eric
Backus's ports of the FSF utils, I consistently got "can't cat input
to itself" or similar from those builds. I think the particular idiom
that caused trouble was something like 'command | cat - file > temp'.
I bet this port would have the same problems, since the 'first'
cluster of stdin or stdout may or may not be available to stat(). In
particular, in the case above since the temp file has been truncated,
it would not. That would mean that the file and text utilities would
not necesarily solve many makefile porting problems :-(.
--
Stephen Turnbull / Yaseppochi-gumi / <turnbull AT shako DOT sk DOT tsukuba DOT ac DOT jp>
http://turnbull.sk.tsukuba.ac.jp/ anon FTP: turnbull.sk.tsukuba.ac.jp
Check out Kansai-WWW, too ------------> http://pclsp2.kuicr.kyoto-u.ac.jp/
- Raw text -