Re: flexfax: HylaFAXServer::getSequenceNumber() shouldn't use O_EXCL

[Darren Nickerson <darren@dazza.org>]

>>>>> On Thu, 16 Mar 2000, "MB" == Mark Burton wrote:

  MB> I don't currently subscribe to the mailing list so I apologise if this
  MB> has been covered 50 times already.

Unlikely for the CVS to be discussed much on this list. That belongs to 
another list. Before spending much time with the CVS you really should join 
the appropriate lists, or at the very least have a look at the archives, no? Do
n't get me wrong . . . the more people we have banging on the CVS the better 
at this point, we really need to hammer out a release!

  MB> I have installed a recent version of hylafax 4.1beta2 (from the CVS
  MB> repository) on my Linux box (2.2.14) and a problem I had to fix was
  MB> that sends were failing with messages like this:

  MB> Mar 16 13:53:00 rook HylaFAX[14565]: docq/seqf: open: File exists

  MB> It seems to me that the code in HylaFAXServer::getSequenceNumber()
  MB> can't possibly work as it combines O_EXCL and O_CREAT open modes which
  MB> will cause the open to fail if the file exists.

  MB> I have therefore removed the O_EXCL and it seems to work OK.

Absolutely correct. There's a more robust fix though, since this code was 
originally broken trying to address a security advisory ;-) That change was 
checked in by Robert a few days ago . . . I've tested it and it seems to do 
the biz.

  MB> One other thing, the files that are created in the docq directory are
  MB> being created with a owner id of uucp (good) and a group id of 60002
  MB> (maybe not good). Is this how it should be?

I'm actually not sure if this is intentional or not. Seems odd, I agree.

-Darren