Path: utzoo!utgpu!water!watmath!clyde!cbosgd!ihnp4!ttrdc!levy From: levy@ttrdc.UUCP (Daniel R. Levy) Newsgroups: comp.unix.wizards Subject: Re: Non-writes to /tmp Message-ID: <2036@ttrdc.UUCP> Date: 13 Dec 87 09:40:28 GMT References: <10776@brl-adm.ARPA> <9513@tekecs.TEK.COM> Organization: AT&T, Skokie, IL Lines: 16 In article <9513@tekecs.TEK.COM>, andrew@frip.gwd.tek.com (Andrew Klossner) writes: >On my system, the major use of /tmp files is for one phase of a >compiler to pass information to the next. The close in the first >(writing) phase seems to force the blocks to disk. Or does it? No, the blocks haven't necessarily gone to disk yet even on close. The blocks will go to disk in a short time thereafter, though, the time depending on the system and how it was set up. Some systems depend upon an /etc/update (or similar) daemon, which invokes the sync() system call on a periodic basis to cause buffered blocks to be written to disk. Others (like System V) have this periodic update functionality built into the kernel. -- |------------Dan Levy------------| Path: ..!{akgua,homxb,ihnp4,ltuxa,mvuxa, | an Engihacker @ |}!ttrdc!ttrda!levy | AT&T Computer Systems Division | Disclaimer? Huh? What disclaimer??? |--------Skokie, Illinois--------|