Re: LYNX-DEV Improvement on /tmp code fix?

[Jonathan Sergent]

Tom wrote:
 ] sticky directories are fairly recent - I don't remember noticing them before
 ] Sun's NFS stuff took hold in the early 90's. symbolic links have been around
 ] a bit longer than that (they weren't in SVr1, I think).

Didn't 4.3BSD have them?  That's been around at least 10 years.  
en.ecn.purdue.edu had them (past tense, it [a Gould NP1] got turned 
off a month or so ago) from what I remember.  Could've been a ghg
hack, though...

 ] if you open the file with the right permissions, you don't need to chmod it.
 ] you should be able to wrapper this to construct the filename and return the
 ] pointer to it as one operation, that way you aren't going to be concerned
 ] about someone else using the file.

Desire to chmod was not related to desire to avoid symlink spoofing, it 
was related to desire for concern for privacy issues.  umask() would
work fine, but if you have three-argument open then you have to put
_something_ in there.

 ] that's expensive, and you have to know how to transit mount points (that's
 ] not very portable either - it'd be a pain to implement)

Fine, we just trust admins on that part then?

I envisioned just splitting the path into directory components and statting
each one, since we have a full path to temp_space.

I'll look into the code that Jim Dennis (saw you in July BYTE page 16) 
posted a reference to...


-- 
Jonathan Sergent / address@hidden / address@hidden
;
; To UNSUBSCRIBE:  Send a mail message to address@hidden
;                  with "unsubscribe lynx-dev" (without the
;                  quotation marks) on a line by itself.
;