Number: 1132

Date: 18-May-84 12':32':32



Subject: Race condition on INFILEP?

Lisp Version: 14-May-84 21':12':25

Description: TEdit was taking a long time to format a press file, so I started up another process to watch for the appearance of the file, and then empress it; something like'
but the INFILEP succeeded while TEDIT still was writing it , and when the LISTFILES started to run, it got a leaf error "file busy in conflicting way".  Waiting a few seconds more "by hand" allowed the next call to LISTFILES to win.'
Date': 22 May 84 15':26 PDT'
Declined.  There is no race.  INFILEP does not mean "can I open this file for input?", and the manual even discusses this aspect.  Some file systems (eg. IFS''s, unfortunately) are happy to claim a file exists before someone has closed it.  Perhaps you wanted (until (NLSETQ (OPENFILE & ''INPUT)) --).'
Date': 22 May 84 17':45 PDT'
Subject': Re': AR#1132': Race condition on INFILEP?'
In-reply-to':''s message of 22 May 84 15':26 PDT'
Foo, this is clearly a bug and not one of the cases discussed in the manual, which at best suggests lock-out by other users/writers at the file server (I was locked-out against "myself").'
Even if one considers this behaviour a "feature", he would still want a variant of the rejected "can I open this file for input?" primitive; perhaps a good way to achieve this is to permit a NOERROR option to OPENFILE.'
-- JonL --'


Test Case: 

Edit-By: Sannella.PA

Edit-Date: 24-May-84 16':24':31


Assigned To: 



System: Operating System

Subsystem: Generic File Operations

Machine: 1132


Microcode Version: 5124

Memory Size: 4096

File Server: 

Server Software Version: 

Difficulty: Hard

Frequency: Intermittent

Impact: Serious

Priority: Hopefully

Status: Declined

Problem Type: Bug

Source Files: