home *** CD-ROM | disk | FTP | other *** search
- Submitted-by: djm@eng.umd.edu (David J. MacKenzie)
-
- > The function getcwd() is in most cases a passive function, except
- > in one (and perhaps others): SunOS 4.1.1.
-
- > This system, as indicated in the man pages clearly states that
- > getcwd simply makes a system("pwd") to get the result.
-
- Sun merely picked it up from SVR3, which does the same thing.
- Actually, I bet it uses popen, but regardless, it does spawn a shell.
- What an obscene implementation! I use Doug Gwyn's PD version when I
- compile things; it doesn't start any extra processes or mess with signals.
- On BSD-like systems, such as Suns, I usually use getwd instead of getwd.
- That also doesn't start any processes.
-
- > Maybe it would be worth to add description about how the signals should
- > behave when calling standardized functions. I checked the draft against
- > any particular information on this, but it does not specify that
- > getcwd() should not alter signals.
-
- Maybe they didn't mention it on purpose, to allow the existing SVR3/SunOS
- implementation to conform. POSIX often makes concessions to existing
- practice by leaving things vague or unspecified. (This is one way
- that it sometimes handles historical SysV/BSD differences.)
-
-
- Volume-Number: Volume 28, Number 11
-
-