engine tests ignored since Aug 2019

[DanGrayson]

The tests in M2/Macaulay2/tests/engine are not run (by "make") because "engine" is not listed as a subdirectory in M2/Macaulay2/tests/Makefile.in. That's been the case since August, 2019. Wait for #1190, as it repairs the tests, which since then have been broken.

[mahrud]

I repaired those that I could understand and fix easily, but not all, and some time out, which could be regressions but I don't know.

• goals/ISSAC-97.m2 (Failed)
• goals/gbZZ.m2 (Failed)
• goals/joswig.m2 (Timeout)
• quarantine/2-homog-bug.m2 (Failed)
• quarantine/C05.m2 (Failed)
• quarantine/HH.m2 (Failed)
• quarantine/issac-97.m2 (Failed) (added in PrimaryDecomposition, MinimalPrimes, Saturation #1385)
• quarantine/ker8.m2 (Failed)
• quarantine/lapack.m2 (Failed)
• quarantine/newlines.m2 (Failed)
• quarantine/testmulti.m2 (Failed)
• engine/gb-bench.m2 (Timeout) (closed in PrimaryDecomposition, MinimalPrimes, Saturation #1385)
• engine/normal.m2 (Failed)
• engine/raw-decompose.m2 (Failed)
• engine/raw-freemodule.m2 (Failed)
• engine/raw-monomial.m2 (Failed)
• engine/raw-numerics.m2 (Failed)
• engine/raw-res.m2 (Failed)
• engine/raw-ring.m2 (Failed); two remain to close equality of raw rings #1086:
  • assert(raw RR_53 === rawRR(53))
  • assert(raw CC_53 === rawCC(53))
• engine/raw-ring2.m2 (Failed)
• engine/raw-ring3.m2 (Failed)
• engine/raw-subduction.m2 (Failed)
• engine/raw.m2 (Failed)
• engine/raw3.m2 (Failed)
• engine/test-linalg.m2 (Failed)

[DanGrayson]

@mikestillman -- can the rest be repaired?

[mahrud]

If you build with cmake, you can test these with ctest -R engine, by the way. More options here.

[mahrud]

Pinning this, because fixing each of those failed tests is a good issue to resolve.

[DanGrayson]

What does "pinning" mean?

[mahrud]

By the way, there are some notes in the file m2/workingnotes.m2 related to this:

M2/M2/Macaulay2/m2/workingnotes.m2

Lines 3 to 11 in 3a00dcf

	test/ISSAC-97.m2 -- too big now, improve the algorithm
	test/engine-div.m2 -- raw % and // need to be defined, or something
	test/gbZZbug.m2 -- gb problem with bit integers
	test/galois.m2 -- GF substitution
	test/subst3.m2 -- GF substitution
	test/euclid.errors -- % and // in ZZ[x,Inverses=>true,MonomialOrder=>RevLex]
	test/engine/LU.m2 -- in debug version, I get many smashed blocks! Memory allocator problem?
	test/order.m2 -- sorting of polynomials needs a total ordering
	test/slow/completeintersections.m2 -- too slow or hung up???

That file hasn't changed since 2006. Probably could be safely removed? In fact, none of these M2 sources seem to be used anywhere:

• Core.m2: only has an error line
• macaulay1.m2: contains only 2 functions
• overview.m2: seems to contain a bunch of documentation nodes that are not loaded
• sagbi.m2: is this in working condition?
• version.m2.in
• workingnotes.m2

There are also some none M2 sources that don't seem to used:

• TAGS.doc: seems to be empty
• Makefile.tests: not used anywhere
• phases.mak: windows related
• COPYRIGHT: every file in this directory already has a copyright header

[mahrud]

Quoting from a comment in #1387:

For engine/raw_res.m2, the problem appears to be in the printing of a RawComputation. The example that fails is real projective plane, algorithm 0. If we add semicolons to the end of lines 123, 124, and 125 (which suppresses printing the object returned by rawResolution, rawGBSetStop, and rawStartComputation) the assertions pass. The actual error message that appears can be found in the function BettiDisplay::entry, which is line 71 in e/betti.cpp, but we're not sure how control passes to this function or why the object is in a bad state when this function is called. Any suggestions?

Originally posted by @dylanpeifer in #1387 (comment)