From: Craig A Summerhill (craig@cni.org)
Date: Thu Apr 22 1999 - 05:01:10 CDT
On Thu, 22 Apr 1999, John Finlay <finlay@moeraki.com> wrote:
>
> Daniel Stenberg wrote:
> >
> > No, that particular fix doesn't. Hypermail already checks for
> > existing file names before it saves an attachment. It then tries
> > simple alterations (by prepending a single letter to the filename,
> > a to z) and if none succeeds it falls back to generating a random one.
>
> Rather than generating a random name it would seem more useful to
> prepend the attachment name with the message number though I suppose
> one message could have attachements with the same file name.
I realize, to some degree, the likelihood of an attachment using the
same name is a function of the size (and nature) of an archive...
nevertheless, I actually feel this function as it exists in the
current code is sufficient.
I discovered the fact that hypermail would do this by mistake when I
was doing some testing. I was overwriting entire archives without
cleaning the target directory out and discovered I had a number of
MIME attachments which were nearly identical in name (file.xcl,
filea.xcl, fileb,xcl, etc.).
Since then, I have converted tens (perhaps hundreds) of thousands
of messages in old archives and have yet to experience one instance
where hypermail had to alter the name to make it unique. I don't
think I've got a single example of hypermail being forced to add
a character to create a unique filename...
-- Craig A. Summerhill, Systems Coordinator and Program Officer Coalition for Networked Information 21 Dupont Circle, N.W., Washington, D.C. 20036 Internet: craig@cni.org AT&Tnet (202) 296-5098