| To: | yyyyyyyyyyyyyyy@xxxxxxxxxxxxx |
|---|---|
| Subject: | Re: Defect in XBD 3.382_System_Console |
| From: | yyyyyyyyyy@xxxxxxx |
| Date: | Fri, 8 Aug 2003 13:14:09 EDT |
In a message dated 2003-08-07 9:03:34 PM Eastern Daylight Time, Larry.Dwyer@xxxxxx writes:The user of fmtmsg(..) or syslog(..) doesn't know how many characters may precede the message, so it is not possible to reliably determine how long to make a log message. Are you suggesting that we should mandate a minimum line length for the system console? If the minimum set of characteristics includes the line length, then I will report that the minimum length of the line to my console is 0 (because my POSIX conforming system does not have a display connected to my "console"). On the other hand, the length might depend on what hardware the customer connects to the console spigot. Actually, fmtmsg() presumes that a console and stderr, however present, allows a minimum line length of 25 chars to handle the maximum width label field without truncation. syslog() requires it be able to show a text representation of a PID and a time stamp, and presumes a system console is always available to handle the LOG_CONS option - "ENOSYS: Console not found" is not an allowed error return. In terms of physical implementation, a display doesn't have to be connected, I agree; it could be a file for perusal later fullscreen or in a subwindow of the main application terminal, or a network pipe to a separate system, but from a logical standpoint the console behaves as a terminal type device of varying capabilities, some of which may make it suitable for other uses than just error reporting. The standard does have various enviroment variables and the termio structures to document the characteristics of the application consoles, but it sems the details of such description, if the implementation supports separate hardware, for a system console were glossed over at some point, leaving just the "implementation-defined" in the XBD definition. I would also take these characteristics to include the maximum widths of the optional elements that fmtmsg() and syslog() do produce so that an application can be tailored to function well on that implementation. Once more, Mark |
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| ||
| Previous by Date: | Re: Defect in XBD 3.386_System_Reboot, SHwareSyst |
|---|---|
| Next by Date: | Defect in XCU c99, don.cragun |
| Previous by Thread: | Re: Defect in XBD 3.382_System_Console, Larry Dwyer |
| Next by Thread: | Defect in XBD 3.385_System_Process, ajosey |
| Indexes: | [Date] [Thread] [All Lists] |