wxmaxima freezes running test suite

[beew]

Wxmaxima freezes for run_testsuite(display_all = true) , the top command shows that wxmaxima is consuming 100% cpu (in addition to 100% from sbcl)

then at problem 80 (line 232), it looks like it just quits. Top doesn't show any cpu activity even though wxmaxima is still running (just that the test session is killed apparently)

Run_testsuite() without display works fine and all the cpu consumption is from sbcl, wxmaxima is taking up less than 1%.

run_testsuite(display_all = true) works without issue in xmaxima, so it is not a maxima issue.

OS Ubuntu 22.04, wxmaxima 23.02.0, maxima 5.45.1

[daute]

Thanks for the report.
run_testsuite(display_all=true, tests=[rtestode]);
creates a XML error with an up-to-date Maxima (compiled using ECL). Will check in the next days, what's going on here...

[gunterkoenigsmann]

I tried the whole thing using a maxima that was compiles with SBCL and with both maxima and wxMaxima from the current git HEAD. In that case wxMaxima is busy for hours layouting all the math maxima produces. But it eventually finishes without any XML error, works fine afterwards, but still requires half of a CPU on my system. I guess I do want to profile this once the current bugs are resolved. My maxima-ECL won't start currently, and --logtostderrdoesn't work for me => seems we found an excellent, even if very long-running test case.

[daute]

The following code from the testsuite seems problematic:

''(f(x) = sin(x)*(at('diff(f(x),x,1),x = 0)-2)+f(0)*cos(x)+2*x);

/* BUG 3899 */
errcatch(desolve('diff(f(x),x),x));

This results in the following XML (as seen in the XML monitor):

<math>
  <t>desolve(<eqn>,<depvar>): dependant variable <depvar> cannot be a mapatom, found: </t>
     <mi>x</mi>
     </math>

Which of course is not valid XML, the "tag" <depvar> is closed by </t>. There must be a replacement with the corresponding entitites. Somewhere in our Lisp code...

This is found by the following code:
src/wxMaxima.cpp(1906): assert "cell" failed in DoConsoleAppend(): There was an error in generated XML!

[gunterkoenigsmann]

Seems we lack a (wxxml-fix-string ) somewhere that escapes the < and > before sending them to Maxima. We'll have to search for where the <t> flags are generated...

[gunterkoenigsmann]

Yesterday's release should allow the user interface more often to send signals even when wxMaxima is busy - and nearly all of the XML bugs I found have been resolved in the meantime => I would say that bug report has greatly improved the quality of wxMaxima. Let's hope I managed to resolve the actual issue, as well.

[beew]

Yesterday's release should allow the user interface more often to send signals even when wxMaxima is busy - and nearly all of the XML bugs I found have been resolved in the meantime => I would say that bug report has greatly improved the quality of wxMaxima. Let's hope I managed to resolve the actual issue, as well.

Still happening in wxmaxima 23.04.1. Wxmaxima is still using 100% cpu.