summaryrefslogtreecommitdiffstats
path: root/win/tclWinPipe.c
Commit message (Collapse)AuthorAgeFilesLines
* * win/tclWinPipe.c (TclpCreateProcess): change panics to Tclhobbs2006-03-291-31/+38
| | | | errors and do proper refcounting of noe objPtr. [bug 1194429]
* * doc/CrtChannel.3: Added TCL_CHANNEL_VERSION_5, made itandreas_kupries2006-03-271-2/+3
| | | | | | | | | | | | | | * generic/tcl.h: the version where the "truncateProc" * generic/tclIO.c: is defined at, and moved all channel * generic/tclIOGT.c: drivers of Tcl to v5. * generic/tclIORChan.c: * unix/tclUnixChan.c: * unix/tclUnixPipe.c: * win/tclWinChan.c: * win/tclWinConsole.c: * win/tclWinPipe.c: * win/tclWinSerial.c: * win/tclWinSock.c:
* Finalization of the sockets/pipesvasiljevic2006-03-101-33/+10
| | | | | | | | is now solely done in TclpFinalizeSockets() and TclpFinalizePipes() and not over the thread-exit handler, because the order of actions the Tcl generic core will impose may result in cores/hangs if the thread exit handler tears down corresponding subsystem(s) too early.
* * win/tclWinPort.h: Applied patch #1267871 by Matt Newman forpatthoyts2005-11-041-9/+9
| | | | | | * win/tclWinPipe.c: extended error code support on Windows. * tests/exec.test: Tests for extended error codes. * generic/tclPipe.c: Permit long codes (platform macros permitting).
* ANSIfydkf2005-11-041-28/+30
|
* * generic/tclPipe.c (TclCreatePipeline): Arrange for POSIX systems todkf2005-07-281-2/+2
| | | | | | | | * unix/tclUnixPipe.c (TclpOpenFile): use the O_APPEND flag for * tests/exec.test (exec-19.1): files opened in a pipeline like ">>this". Note that Windows cannot support such access; there is no equivalent flag on the handle that can be set at the kernel-call level. The test is unix-specific in every way. [Bug 1245953]
* Getting more systematic about styledkf2005-07-241-647/+642
|
* bug 1225044Kevin B Kenny2005-06-221-2/+7
|
* bug 1225727Kevin B Kenny2005-06-221-7/+4
|
* Merged kennykb-numerics-branch back to the head; TIPs 132 and 232Kevin B Kenny2005-05-101-1/+1
|
* TIP#218 IMPLEMENTATIONandreas_kupries2005-01-271-4/+63
| | | | | | | | | | | | | | | | | | | | * generic/tclDecls.h: Regenerated from tcl.decls. * generic/tclStubInit.c: * doc/CrtChannel.3: Documentation of extended API, * generic/tcl.decls: extended testsuite, and * generic/tcl.h: implementation. Removal of old * generic/tclIO.c: driver-specific TclpCut/Splice * generic/tclInt.h: functions. Replaced with generic * tests/io.test: thread-action calls through the * unix/tclUnixChan.c: new hooks. Update of all builtin * unix/tclUnixPipe.c: channel drivers to version 4. * unix/tclUnixSock.c: Windows drivers extended to * win/tclWinChan.c: manage thread state in a thread * win/tclWinConsole.c: action handler. * win/tclWinPipe.c: * win/tclWinSerial.c: * win/tclWinSock.c:
* * generic/tclUtil.c: Updated Tcl_GetNameOfExecutable() todgp2004-12-011-6/+2
| | | | | | | | | | | | | | | | | | | | | | | | * generic/tclEncoding.c: make use of a ProcessGlobalValue for * generic/tclEvent.c: storing the executable name. Added internal routines Tcl(Get|Set)ObjNameOfExecutable() to access that storage in Tcl_Obj, rather than string format. * unix/tclUnixFile.c: Rewrote TclpFindExecutable() to use * win/tclWinFile.c: TclSetObjNameOfExecutable to store the executable name it computes. * generic/tclInt.h: Added internal stub entries for * generic/tclInt.decls: TclpFindExecutable and Tcl(Get|Set)ObjNameOfExecutable. * generic/tclIntDecls.h: make genstubs * generic/tclStubInit.c: * generic/tclCmdIL.c: Retrieve executable name in Tcl_Obj form * win/tclWinPipe.c: instead of string form. * unix/tclUnixTest.c: Update [testfindexecutable] command to use new internal interfaces.
* Patch 976520 reworks several of the details involved withdgp2004-11-301-2/+6
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | startup/initialization of the Tcl library, focused on the activities of Tcl_FindExecutable(). * generic/tclIO.c: Removed bogus claim in comment that encoding "iso8859-1" is "built-in" to Tcl. * generic/tclInt.h: Created a new struct ProcessGlobalValue, * generic/tclUtil.c: routines Tcl(Get|Set)ProcessGlobalValue, and function type TclInitProcessGlobalValueProc. Together, these take care of the housekeeping for "values" (things that can be held in a Tcl_Obj) that are global across a whole process. That is, they are shared among multiple threads, and epoch and mutex protection must govern the validity of cached copies maintained in each thread. * generic/tclNotify.c: Modified TclInitNotifier() to tolerate being called multiple times in the same thread. * generic/tclEvent.c: Dropped the unused argv0 argument to TclInitSubsystems(). Removed machinery to unsure only one TclInitNotifier() call per thread, now that that is safe. Converted Tcl(Get|Set)LibraryPath to use a ProcessGlobalValue, and moved them to tclEncoding.c. * generic/tclBasic.c: Updated caller. * generic/tclInt.h: TclpFindExecutable now returns void. * unix/tclUnixFile.c: * win/tclWinFile.c: * win/tclWinPipe.c: * generic/tclEncoding.c: Built new encoding search initialization on a foundation of ProcessGlobalValues, exposing new routines Tcl(Get|Set)EncodingSearchPath. A cache of a map from encoding name to directory pathname keeps track of where encodings are available for loading. Tcl_FindExecutable greatly simplified into just three function calls. The "library path" is now misnamed, as its only remaining purpose is as a foundation for the default encoding search path. * generic/tclInterp.c: Inlined the initScript that is evaluated by Tcl_Init(). Added verification after initScript evaluation that Tcl can find its installed *.enc files, and that it has initialized [encoding system] in agreement with what the environment expects. [tclInit] no longer driven by the value of $::tcl_libPath; it largely constructs its own search path now, rather than attempt to share one with the encoding system. * unix/tclUnixInit.c: TclpSetInitialEncodings factored so that a new * win/tclWinInit.c: routine TclpGetEncodingNameFromEnvironment can reveal that Tcl thinks the [encoding system] should be, even when an incomplete encoding search path, or a missing *.enc file won't allow that initialization to succeed. TclpInitLibraryPath reworked as an initializer of a ProcessGlobalValue. * unix/tclUnixTest.c: Update implementations of [testfindexecutable], [testgetdefenc], and [testsetdefenc]. * tests/unixInit.test: Corrected tests to operate properly even when a value of TCL_LIBRARY is required to find encodings. * generic/tclInt.decls: New internal stubs: TclGetEncodingSearchPath, TclSetEncodingSearchPath, TclpGetEncodingNameFromEnvironment. These are candidates for public exposure by future TIPs. * generic/tclIntDecls.h: make genstubs * generic/tclStubInit.c: * generic/tclTest.c: Updated [testencoding] to use * tests/encoding.test: Tcl(Get|Set)EncodingSearchPath. Updated tests.
* * win/tclWinPipe.c: The pipe channel driver now respects the -blocking optiondavygrvy2004-11-091-47/+72
| | | | | | | | | | when closing is the same way the UNIX side works. This is avoid a hung shell when exiting due to open pipes that refuse to close in a graceful manner. ***POTENTIAL INCOMPATIBILITY*** Scripts that use async pipes on windows, must (like the UNIX side) set -blocking to 1 before calling [close] to receive the exit status. [Bug 947693]
* * generic/tclBasic.c:dgp2004-10-061-4/+4
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | * generic/tclBinary.c: * generic/tclCmdAH.c: * generic/tclCmdIL.c: * generic/tclCmdMZ.c: * generic/tclCompExpr.c: * generic/tclDictObj.c: * generic/tclEncoding.c: * generic/tclExecute.c: * generic/tclFCmd.c: * generic/tclHistory.c: * generic/tclIndexObj.c: * generic/tclInterp.c: * generic/tclIO.c: * generic/tclIOCmd.c: * generic/tclNamesp.c: * generic/tclObj.c: * generic/tclPkg.c: * generic/tclResult.c: * generic/tclScan.c: * generic/tclTimer.c: * generic/tclTrace.c: * generic/tclUtil.c: * generic/tclVar.c: * unix/tclUnixFCmd.c: * unix/tclUnixPipe.c: * win/tclWinDde.c: * win/tclWinFCmd.c: * win/tclWinPipe.c: It is a poor practice to directly set or append to the value of the objResult of an interp, because that value might be shared, and in that circumstance a Tcl_Panic() will be the result. Searched for example of this practice and replaced with safer alternatives, often using the Tcl_AppendResult() routine that dkf just rehabilitated. * library/dde/pkgIndex.tcl: Bump to dde 1.3.1
* More formatting fixes...dkf2004-10-061-20/+18
|
* * win/tclWinPipe.c:davygrvy2004-05-301-44/+56
| | | | | | | | | * win/tclWinPort.h: Reworked the win implementation of Tcl_WaitPid to support exitcodes in the 'signed short' range. Even though this range is non-portable, it is valid on windows. Detection of exception codes are now more accurate. Previously, an application that exited with ExitProcess((DWORD)-1); was improperly reported as exiting with SIGABRT.
* (BuildCommandLine): Use the existence of contents in the linePtr ratherdavygrvy2004-05-101-2/+2
| | | | than the scratch DString post the append, as this more clear.
* (TclpCreateProcess): When under NT, with no console, and executing adavygrvy2004-05-101-2/+2
| | | | | DOS application, the path priming does not need an ending space as BuildCommandLine() will do this for us.
* (BuildCommandLine): Moved non-obvious appending logic to outside the loopdavygrvy2004-05-101-6/+7
| | | | and added commentary for its purpose.
* Refix bug 789040 re-entered in rev 1.41chengyemao2004-05-081-2/+3
|
* * tests/winPipe.test: Six more cases added.davygrvy2004-02-021-6/+2
| | | | | * win/tclWinPipe.c: Fixed BuildCommandLine() to pass the new cases.
* * tests/winPipe.test: Added proof that BuildCommandLine() is notdavygrvy2004-02-021-8/+12
| | | | | | | | doing the "N backslashes followed a quote -> insert N * 2 + 1 backslashes then a quote" rule needed for the crt's parse_cmdline(). * win/tclWinPipe.c: Fixed BuildCommandLine() to pass the new cases.
* * tests/winPipe.test: more pass-thru commandline verifications.davygrvy2004-02-011-8/+2
| | | | | | | * win/tclWinPipe.c (BuildCommandLine): Special case quoting for '{' not required by the c-runtimes's _setargv(). * win/tclAppInit.c: Removed our custom setargv() in favor of the one provided by the c-runtime. [Bug 672938]
* * win/tclWinPipe.c (Tcl_WaitPid): Fixed a thread-safety problem with thedavygrvy2004-01-201-5/+14
| | | | | | process list. The delayed cut operation after the wait was going stale by being outside the list lock. It now cuts within the lock and does a locked splice for when it needs to instead. [Bug 859820]
* All uses of 'panic' (the macro) changeddavygrvy2003-12-241-7/+9
| | | | | | | to 'Tcl_Panic' (the function). The #define of panic in tcl.h clearly states it is deprecated in the comments. [Patch 865264]
* * win/tclWinPipe.c (BuildCommandLine): Applied the patch comingandreas_kupries2003-10-211-3/+6
| | | | | with [Bug 805605] to the code, fixing the incorrect use of ispace noted by Ronald Dauster <ronaldd@users.sourceforge.net>.
* * win/tclWinPipe.c: The windows port of expect can calldavygrvy2003-09-281-1/+4
| | | | | | TclWinAddProcess before any of the other pipe functions. Added a missing PipeInit() call to make sure the initialization happens.
* Bug fix in BuildCommandLine: missing a space when linePtr is not nullchengyemao2003-08-151-2/+3
|
* * win/Makefile.in: Don't define TCL_DBGXmdejong2003-04-151-7/+2
| | | | | | | | | | | | | | | symbol for every compile. Instead, define TCL_PIPE_DLL only when compiling tclWinPipe.c. This will break other build systems, so they will need to remove the TCL_DBGX define and replace it with a define for TCL_PIPE_DLL. * win/tclWinPipe.c (TclpCreateProcess): Remove PREFIX_IDENT and DEBUG_IDENT from top of file. Use TCL_PIPE_DLL passed in from build env instead of trying to construct the dll name from already defined symbols. This approach is more flexible and better in the long run.
* * generic/tcl.h: Removed TCL_PREFIX_IDENT and TCL_DEBUG_IDENTdgp2003-03-121-3/+6
| | | | | * win/tclWinPipe.c: from tcl.h -- they are not part of Tcl's public interface. Put them in win/tclWinPipe.c where they are used.
* * generic/tcl.h: Add TCL_PREFIX_IDENT andmdejong2003-01-281-3/+35
| | | | | | | | | | | | | | | | | TCL_DEBUG_IDENT, used only by TclpCreateProcess. * unix/Makefile.in: Define TCL_DBGX. * win/Makefile.in: Define TCL_DBGX. * win/tclWinPipe.c (TclpCreateProcess): Check that the Tcl pipe dll actually exists in the Tcl bin directory and panic if it is not found. Incorporate TCL_DBGX into the Tcl pipe dll name. This fixes a really mysterious error that would show up when exec'ing a 16 bit application under Win95 or Win98 when Tcl was compiled with symbols. The error seemed to indicate that the executable could not be found, but it was actually the Tcl pipe dll that could not be found.
* * generic/tclPipe.c (TclCleanupChildren):davygrvy2002-12-171-5/+52
| | | | | | | | | | | | | | * tests/winPipe.c: * win/tclWinPipe.c (Tcl_WaitPid): * win/tclWinTest.c: Gave Tcl_WaitPid the ability to return a Win32 exception code translated into a posix style SIG*. This allows [close] to report "CHILDKILLED" without the meaning getting lost in a truncated exit code. In TclCleanupChildren(), TclpGetPid() had to get moved to before Tcl_WaitPid() as the the handle is removed from the list taking away the ability to get the process id after the wait is done. This shouldn't effect the unix implimentaion unless waitpid is called with a pid of zero, meaning "any". I don't think it is..
* * win/tclWinPipe.c: reverted back to -r1.27 due to numerous testdavygrvy2002-12-051-67/+7
| | | | failures that need to be resolved first.
* * win/tclWinPipe.c (Tcl_WaitPid): When a process exits with andavygrvy2002-12-041-5/+52
| | | | | | | | | | | | | exception, pass this notice on to the caller with a SIG* code rather than truncating the exit code and missing the meaning. This allows TclCleanupChildren() to report "CHILDKILLED". This has a different behavior than unix in that closing the read pipe to a process sends the SIGPIPE signal which is returned as a SIGPIPE exit status. On windows, we send the process a CTRL_BREAK_EVENT and get back a CONTROL_C_EXIT which is documented to mean a SIGINT which seems wrong as a system, but is the correct exit status.
* * win/tclWinPipe.c (PipeClose2Proc): Changed CTRL_C_EVENTdavygrvy2002-12-041-2/+2
| | | | to CTRL_BREAK_EVENT as it can't be ignored by the child.
* * win/tclAppInit.c (sigHandler): Protect from trying to close adavygrvy2002-12-041-3/+16
| | | | | | | | | | | | | | | | | | NULL handle. * win/tclWinPipe.c (PipeClose2Proc, TclpCreateProcess): Send a real Win32 signal (CTRL_C_EVENT) when the read channel is brought down to alert the child to close on its side. Start the process with CREATE_NEW_PROCESS_GROUP to allow the ability to send these signals. The following test case now brings down the child without the use of an external [kill] command. % set p [open "|[info name]" w+] file8d5380 % pid $p 2876 % close $p <- now doesn't block in Tcl_WaitPid() %
* * win/tclWinConsole.c:davygrvy2002-11-261-37/+75
| | | | | | | | | | | | | | | | * win/tclWinPipe.c: * win/tclWinSerial.c: * win/tclWinSock.c: * win/tclWinThrd.c: * win/tclWinTime.c: General cleanup of all worker threads used by the channel drivers. Eliminates the normal case where the worker thread is terminated ('cept the winsock one). Instead, use kernel events to signal a clean exit. Only when the worker thread is blocked on an I/O call is the thread terminated. Essentially, this makes all other channel worker threads behave like the PipeReaderThread() function for it's cleaner exit behavior. This appears to fix [Bug 597924] but needs 3rd party confirmation to close the issue.
* * generic/tclEvent.c (TclInExit, TclInThreadExit):mdejong2002-11-071-4/+4
| | | | | | | | | | | | | | | Split out functionality of TclInExit to make it clear which one should be called in each situation. * generic/tclInt.decls: Declare TclInThreadExit. * generic/tclIntDecls.h: Regen. * generic/tclStubInit.c: Regen. * mac/tclMacChan.c (StdIOClose): * unix/tclUnixChan.c (FileCloseProc): * win/tclWinChan.c (FileCloseProc): * win/tclWinConsole.c (ConsoleCloseProc): * win/tclWinPipe.c (TclpCloseFile): * win/tclWinSerial.c (SerialCloseProc): Invoke the new TclInThreadExit method instead of TclInExit.
* * win/tclWinPipe.c (BuildCommandLine): Fixed bug #554068 ([exec]andreas_kupries2002-06-171-1/+6
| | | | | | on windows did not treat { in filenames well.). Bug reported by Vince Darley <vincentdarley@users.sourceforge.net>, patch provided by Vince too.
* 4 fs fixesvincentdarley2002-03-241-2/+2
|
* * [Patch 505630] Updated interfaces of generic/tclBasic.cc (TIP 27).dgp2002-01-251-3/+3
| | | | | | | * [Patch 506818] Updated interfaces of generic/tclHash.c (TIP 27). * [Patch 506807] Updated interfaces of generic/tclObj.c (TIP 27). * [Patch 507304] Updated interfaces of win/tclWin32.c (TIP 27). * Update all callers.
* * Updated interfaces of generic/tclEncoding, generic/tclFilename.c,dgp2002-01-251-4/+4
| | | | | | | | | | | generic/tclIOUtil.c, generic/tclPipe.c, generic/tclResult.c, generic/tclUtil.c, generic/tclVar.c and mac/tclMacResource.c according to TIP 27. Tcl_TranslateFileName rewritten as wrapper around VFS-aware version. Updated callers. ***POTENTIAL INCOMPATIBILITY*** Includes source incompatibilities: argv arguments of Tcl_Concat, Tcl_JoinPath, Tcl_OpenCommandChannel, Tcl_Merge; argvPtr arguments of Tcl_SplitList and Tcl_SplitPath.
* * Updated APIs in the file generic/tclIO.c according to the guidelinesdgp2002-01-151-4/+4
| | | | | | | | | | | of TIP 27. Several minor documentation corrections as well. * Updated channel driver interface according to the guidelines of TIP 27. See also [Bug 500348]. * Moved Tcl_EolTranslation enum declaration from generic/tcl.h to generic/tclInt.h (renamed to TclEolTranslation). It is not used anywhere in Tcl's public interface.
* * win/tclWinPipe.c: More Borland compatibility fixes.davygrvy2001-09-061-2/+5
| | | | [Patch: 436116]
* 2001-08-23 Andreas Kupries <andreas_kupries@users.sourceforge.net>andreas_kupries2001-08-231-3/+3
| | | | | | | | | | * win/tclWinPipe.c (BuildCommandLine): Fixed tcl Bug [432499]. Part of the code used the non-absolute path to the executable to determine quoting. This failed if the absolute path cotained spaces, but the application name itselfl not. This bug caused no trouble on Win NT 5, but does for other variants in the Win* family. Report and fix due to Ken Poole <kenpoole@users.sourceforge.net>.
* Changes from TIP#17 "Redo Tcl's filesystem"vincentdarley2001-07-311-1/+29
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The following files were impacted. * doc/Access.3: * doc/FileSystem.3: * doc/OpenFileChnl.3: * doc/file.n: * doc/glob.n: * generic/tcl.decls: * generic/tcl.h: * generic/tclCmdAH.c: * generic/tclCmdIL.c: * generic/tclCmdMZ.c: * generic/tclDate.c: * generic/tclDecls.h: * generic/tclEncoding.c: * generic/tclFCmd.c: * generic/tclFileName.c: * generic/tclGetDate.y: * generic/tclIO.c: * generic/tclIOCmd.c: * generic/tclIOUtil.c: * generic/tclInt.decls: * generic/tclInt.h: * generic/tclIntDecls.h: * generic/tclLoad.c: * generic/tclStubInit.c: * generic/tclTest.c: * generic/tclUtil.c: * library/init.tcl: * mac/tclMacFCmd.c: * mac/tclMacFile.c: * mac/tclMacInit.c: * mac/tclMacPort.h: * mac/tclMacResource.c: * mac/tclMacTime.c: * tests/cmdAH.test: * tests/event.test: * tests/fCmd.test: * tests/fileName.test: * tests/io.test: * tests/ioCmd.test: * tests/proc-old.test: * tests/registry.test: * tests/unixFCmd.test: * tests/winDde.test: * tests/winFCmd.test: * unix/mkLinks: * unix/tclUnixFCmd.c: * unix/tclUnixFile.c: * unix/tclUnixInit.c: * unix/tclUnixPipe.c: * win/tclWinFCmd.c: * win/tclWinFile.c: * win/tclWinInit.c: * win/tclWinPipe.c
* * win/tclWinPipe.c (PipeClose2Proc): constrained the mutex lock tohobbs2001-07-171-17/+16
| | | | just the TerminateThread call and waiting for termination. (jsmith)
* * win/tclWinConsole.c:mdejong2001-07-161-2/+1
| | | | | | | | | | * win/tclWinPipe.c: * win/tclWinPort.h: * win/tclWinSerial.c: * win/tclWinThrd.c: Remove unnecessary #includes of dos.h, direct.h, and tchar.h. This will help the Cygwin porting effort since these headers do not exist under Cygwin.
* up-port of the stacked channel implementation rewrite in 8.3.2 tohobbs2000-09-281-3/+6
| | | | 8.4a2 code base, merged in with some existing new 8.4a2 features.