>But... my program takes the size returned by lstat and rounds it to the
>nearest 2K, for all types of filesystem objects. This overestimates the
>size by a small amount. Normally not an issue but recently I had a case
>where it created N disks in a case where calculating the sizes correctly
>would have led to N-1. I'm looking for good approximate sizes for the other
>types of objects to improve the accuracy of the program. The mkisofs
>options can't help me here. Well, I could run mkisofs on various samples and
>do statistics, but it seemed to me that someone here might well already know
>the answers...
Non-file objects don't take up any "file" space - if you run isoinfo on
the mkisofs output, symbolic links, device files, pipes etc. have zero size.
Information about non-regular files is stored along with each file's
directory entry - unfortunately there is no easy way of estimating how
large the whole directory will be ... however as each directory entry is
a few 10's of bytes, the total size of the directory entries is likely to be
small in comparison with the whole CD (the value is printed out as
"Total directory bytes: ).
So to answer your question, non-regular files can be ignored - as long as
you allow a suitable "overhead" amount...
James Pearson
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
