I've read some more about perl and forking and the problem seems to be here:
http://perldoc.perl.org/perlfork.html "The open(FOO, "|-") and open(BAR, "-|") constructs are not yet implemented." Unfortunately, of the workarounds offered in that document, only the "pipe_to_fork" works in Windows (tested on ActivePerl 5.8.0 and Strawberry 5.10.0), "pipe_from_fork" does not. I am wondering, though, why forking is used at all - couldn't one use the backtick operator to capture the output from the C preprocessor? The source code im "matwrap" (lines ~170) mentions that this is "likely to get fouled up if any of the arguments had to be quoted[..]. Instead, we want to use exec() with a list argument." But why not backticks? I tried it and it is at least unaffected by quotes. Alternatively, if the "memory penalty" of reading all that output before parsing it is considered a problem, why not in fact use the simple OPEN open command, but quote the execution string with q{}? Thoughts anyone? Daniel On 18.08.2009 20:40, Andrew Ross wrote: > On Tue, Aug 18, 2009 at 05:03:39PM +0200, Daniel Sachse wrote: >> Hi all, >> >> I've sent this question as a Support Request on SFnet and was referred >> to this mailing list (not without some good initial help, thanks smekal). >> >> The case is this: >> >> I'm trying to get PLplot to work together with Octave on a (urks) >> Windows machine. PLplot compiles (and runs) fine for GCC but as soon as >> I enable the Octave binding, I get this: >> >> > mingw32-make >> [..] >> [ 43%] Generating plplot_octave.cc, tmp_stub >> '-' is not recognized as an internal or external command, >> operable program or batch file. >> matwrap: C preprocessor exited with error status >> mingw32-make[2]: *** [bindings/octave/plplot_octave.cc] Error 2 >> mingw32-make[1]: *** [bindings/octave/CMakeFiles/plplot_octave.dir/all] >> Error 2 >> mingw32-make: *** [all] Error 2 >> >> >> The "is not recognized" expression is a Windows command line error >> message when it tries to open/run something that does not exist. >> >> I have traced the error to the perl script matwrap, line ~211 >> (plus/minus) where it seems that the program wants to try and fork - >> which, in Windows, leads to problems. What can I do? >> >> Greetings from Oslo, >> Daniel >> Huge fan of Open Source, but unfortunately bound to Windows. > > Daniel, > > Thank you for your bug report. I don't know whether anyone has tried > octave support for windows before - others can better advise on that. > The octave support uses matwrap to automatically generate the > bindings. Unfortunately this no longer seems to be supported, and > this is at least part of the reason that example 19 is not > implemented - function callbacks are not supported. > > I have been considering trying using swig instead. We already use > swig for generating the java, python and lua bindings. This might > require a bit of work to set up but should in the long run be more > flexible and maintainable since we already have some swig > experience amongst the developers. It should (?) also be portable > to windows. > > Anyway, back to your current problem. I suspect if you had access to > a Linux / Unix box you could generate the necessary files from > matwrap there and then copy them to the build tree on your windows > box. Unless the plplot octave API changes (relatively uncommon) > this will not need to be repeated. Whilst not a perfect solution, > it would be interesting to see if this is enough to get a working > octave binding. > > If you would like to try this, but don't have acccess to a linux > box then let us know and we can probably send you the files > off list. > > Andrew ------------------------------------------------------------------------------ Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day trial. Simplify your report design, integration and deployment - and focus on what you do best, core application coding. Discover what's new with Crystal Reports now. http://p.sf.net/sfu/bobj-july _______________________________________________ Plplot-devel mailing list Plplot-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/plplot-devel
