From: SourceForge.net <no...@so...> - 2006-03-25 03:22:19
|
Bugs item #219161, was opened at 10/25/00 22:02 Message generated for change (Comment added) made by sf-robot You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=110894&aid=219161&group_id=10894 Please note that this message will contain a full copy of the comment thread, including the initial issue submission, for this request, not just the latest update. Category: 27. Channel Types Group: obsolete: 8.0.4 >Status: Closed Resolution: Fixed Priority: 5 Submitted By: Nobody/Anonymous (nobody) Assigned to: Andreas Kupries (andreas_kupries) Summary: Merging stdout and stderr Initial Comment: OriginalBugID: 1470 RFE Version: 8.0.4 SubmitDate: '1999-03-08' LastModified: '2000-04-26' Severity: SER Status: Assigned Submitter: pat ChangedBy: hobbs RelatedBugIDs: 682 582 OS: All FixedDate: '2000-10-25' FixedInVersion: NA ClosedDate: '2000-10-25' Name: Alexandre FERRIEUX DesiredBehavior: It is well known that it's not trivial to merge the stdout and stderr of a child on Windows. The Debated patch by Ian Taylor dragged the existing >@ redirection semantics the wring way (i.e. patching WIndows to mimic UNIX, while for once the UNIX behavior is wrong !!!). Suggestions have been made to use "2>&1" instead, which at least doesn't violate the rest of the >@ semantics. The problem with this is that it's a Bourne shell-based notation, while the rest of [exec/open] uses a csh-notation. In particular , ">&" already exists, which would make "2>&1" at least ambiguous. Hence, I suggest a syntax extension that is more in the spirit of the existing notation, while protecting the semantics: ">&@outpipe". Explanation: ">&@x" already exists, meaning "redirect both stdout and err to the previously opened descriptor found at the low level of the Tcl channel 'x'. Notice that Ian Taylor's patch for Windows and the current (buggy) semantics for Unix both propose to use "stdout" here as "x". Instead, I propose to use a new string "outpipe", which would be recognized internally as 'the output pipe being created'. See 682 for longer explanation of odd behavior. 582 is the "incorrect" patch that Ian Taylor proposed. -- 04/26/2000 hobbs ---------------------------------------------------------------------- >Comment By: SourceForge Robot (sf-robot) Date: 03/24/06 19:22 Message: Logged In: YES user_id=1312539 This Tracker item was closed automatically by the system. It was previously set to a Pending status, and the original submitter did not respond within 14 days (the time period specified by the administrator of this Tracker). ---------------------------------------------------------------------- Comment By: Don Porter (dgp) Date: 03/10/06 08:07 Message: Logged In: YES user_id=80530 Didn't TIP 202 resolve this? ---------------------------------------------------------------------- Comment By: Craig A. Handler (craighandler) Date: 10/15/02 10:11 Message: Logged In: YES user_id=629134 Of course, the difference in behavior Unix v. Windows also occurs with [open "| foo 2>@stdout"]. For this case, at least, the problem could be resolved by supporting "|&" at the head of a new pipeline. This is an already supported syntax in the middle of pipelines, just not at the head of one. So, this should not be objectionable the way 2>&1 seems to be. Of course, it does nothing to resolve the problem for [exec]. ---------------------------------------------------------------------- Comment By: Don Porter (dgp) Date: 07/05/02 13:58 Message: Logged In: YES user_id=80530 This appears to be more FR than Bug. ---------------------------------------------------------------------- You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=110894&aid=219161&group_id=10894 |