Its pretty easy to comment out. Here's probably what is the safest way to do it. Simply go to $XSB_PATH/syslib and copy std_xsb.xwam to std_xsb.xwamsav Then comment out the line
standard_symbol(flush_output,1,file_io).
in std_xsb.P, load xsb, and type
| ?- compile(std_xsb).
This should succeed, and future sessions of xsb will use the modified library file. So just halt and reinvoke.
Terry Swift
On Fri, 1 Jul 2005, Roberto Bagnara wrote:
Terrance Swift wrote:
First, thank you Roberto for your help and patience in finding a bug. I was away from email this morning, which I guess was fortunate for me in that you and David did the work in isolating problematic behavior.
There are actually two issues. The first is that if you call flush_output/1 with an atom that is not an alias it fails silently, rather than raising an exception. This is a bug.
The second is the use of user_output as an alias for the current output stream. I had been under the impression that the ISO standard used user_out rather than user_output, but I should recheck my copy (I dont have it with me for the weekend, and it doesn't seem to be available on line). However, since other Prolog's support user_output (at least Sicstus does) I should also include user_output and user_input as aliases.
They are indeed specified by the ISO standard: 7.10.2.2 and 7.10.2.3, page 55.
I'll make the fixes in sourceforge this afternoon. It may require changes to a couple of files.
As a workaround, you can 1) check out XSB from sourceforge under anonymous CVS (after I make the changes) or 2) use one of these in your code
| ?- flush_output(userout). | ?- flush_output(user_out). | ?- flush_output.
or 3) undefine flush_output/1 as a standard predicate in your copy of XSB (easy, and I can tell you how to do this if this is what you want to do).
I think (3) is the best option for me. In fact, the users involved are unlikely to want to use a non-official release, which rules out (1), while (2) would require me to change the system-independent part of my code. So, if you teach me how to implement (3), I have the temporary workaround I was looking for. Many thanks,
RobertoTerry Swift
On Fri, 1 Jul 2005, Roberto Bagnara wrote:
David Warren wrote:
You've enabled more debugging than I ever have... It gets through the compilation of the emulator, but then fails as it begins to recompile the prolog compiler. It seems to have completed compilation of the first file, so you've gotten reasonably far into the XSB code. THis means it is most likely not the tagging scheme, with which we formerly had problems under fedora.
It looks like here it may have something to do with file i/o? What if you try to execute other simple files with this emulator? I.e., just try to use this emulator to execute something simple.
Going back to the original system, did you configure it (before trying to enable debugging)? Did it get through the recompilation of the prolog compiler and system directories? If so, then this seems to be a problem with the debug configuration. If we go back to the original problem of a loop in some code that ought to run, where was that loop? Could we trace or instrument that original program to try to see where the loop was? Or was it so early, you couldn't find it in the Prolog code?
So, to summarize. It might be the debugging code that is causing the problem here, and it may be easier (or may not) to go back to the original nondebugging version to try to see what's going on there. If it getting into Prolog code, it might be easier to try to track down the loop using Prolog tracing or other Prolog techniques.
Dear David,
you are right: the one I reported is probably a bug concerning the debugging options. The problem was that my code makes heavy use of the foreign language interface and with the optimized version I was unable to understand where the problem occurred.
Now I have configured XSB with
./configure --prefix=/usr/local/lib/xsb --enable-debug --disable-optimization
and this gave me the possibility to ascertain that my problem was in the Prolog part. Everything boils down to the fact that the query
?- flush_output(user_output).silently fails. This is strange, because flush_output/1 is an ISO builtin, user_output is, in ISO Prolog, an alias for the user output stream.
Passing from XSB 2.6 to 2.7.1 I was brought to remove my definition
:- import file_flush/2 from file_io.
flush_output(user_output) :- file_flush(0, _).
because of the error
++Error[XSB]: [Compiler] Trying to redefine the standard predicate: flush_output/1 The file ./xsb_clpq.xwam is removed Aborting...
I thought: "cool, they have incorporated more ISO features!" What am I doing wrong? Many thanks,
Roberto-- Prof. Roberto Bagnara Computer Science Group Department of Mathematics, University of Parma, Italy http://www.cs.unipr.it/~bagnara/ mailto:bagnara@cs.unipr.it
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies from IBM. Find simple to follow Roadmaps, straightforward articles, informative Webcasts and more! Get everything you need to get up to speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click _______________________________________________ Xsb-development mailing list Xsb-development@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/xsb-development
-- Prof. Roberto Bagnara Computer Science Group Department of Mathematics, University of Parma, Italy http://www.cs.unipr.it/~bagnara/ mailto:bagnara@cs.unipr.it