Add a sympy 1.5 run for CircleCI

[eric-wieser]

Relates to #94, although lets start with the last 2.7 release

[codecov]

Codecov Report

Merging #164 into master will increase coverage by 0.08%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master     #164      +/-   ##
==========================================
+ Coverage   67.74%   67.83%   +0.08%     
==========================================
  Files           8        8              
  Lines        4682     4682              
==========================================
+ Hits         3172     3176       +4     
+ Misses       1510     1506       -4

Impacted Files	Coverage Δ
galgebra/mv.py	64.65% <0%> (+0.25%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update d8e533f...cce20f5. Read the comment docs.

[eric-wieser]

One of the failures for examples/ipython/LaTeX.ipynb::Cell 10, after some extra whitespace additions

  \begin{equation*} W
-     = (1/B) \left( X\cdot Y\right)  C \sqrt{\left( X\cdot Y\right) ^{2} - 2 \left( X\cdot Y\right)  \left( X\cdot e\right)  \left( Y\cdot e\right) }
+     =       \left( X\cdot Y\right)  C
-     - (1/B) \left( X\cdot e\right)  \left( Y\cdot e\right)  C \sqrt{\left( X\cdot Y\right) ^{2} - 2 \left( X\cdot Y\right)  \left( X\cdot e\right)  \left( Y\cdot e\right) }
+     -       \left( X\cdot e\right)  \left( Y\cdot e\right)  C
-     + (1/B) \left( X\cdot e\right)  \left( Y\cdot e\right)  \sqrt{\left( X\cdot Y\right) ^{2} - 2 \left( X\cdot Y\right)  \left( X\cdot e\right)  \left( Y\cdot e\right) }
+     +       \left( X\cdot e\right)  \left( Y\cdot e\right)
      + S \sqrt{\left( X\cdot Y\right) ^{2} - 2 \left( X\cdot Y\right)  \left( X\cdot e\right)  \left( Y\cdot e\right) }
  \end{equation*}
  \begin{equation*} \text{Scalar Coefficient}
-     = (1/B) \left( X\cdot e\right)  \left( Y\cdot e\right)  \sqrt{\left( X\cdot Y\right) ^{2} - 2 \left( X\cdot Y\right)  \left( X\cdot e\right)  \left( Y\cdot e\right) }
+     =       \left( X\cdot e\right)  \left( Y\cdot e\right)
  \end{equation*}
  \begin{equation*} \text{Cosh Coefficient}
-     = (1/B) \left( X\cdot Y\right)                          \sqrt{\left( X\cdot Y\right) ^{2} - 2 \left( X\cdot Y\right)  \left( X\cdot e\right)  \left( Y\cdot e\right) }
+     =       \left( X\cdot Y\right)
-     - (1/B) \left( X\cdot e\right)  \left( Y\cdot e\right)  \sqrt{\left( X\cdot Y\right) ^{2} - 2 \left( X\cdot Y\right)  \left( X\cdot e\right)  \left( Y\cdot e\right) }
+     -       \left( X\cdot e\right)  \left( Y\cdot e\right)
  \end{equation*}

Edit: fixed in gh-169

[eric-wieser]

Down to just two failures

[eric-wieser]

One failure!

#x1B[91mNotebook cell execution failed#x1B[0m
#x1B[94mCell 2: Cell outputs differ

Input:
#x1B[0m!python eval_check.py

#x1B[94mTraceback:#x1B[0m
#x1B[94m mismatch 'stdout'#x1B[91m

 assert reference_output == test_output failed:

  'Frame = (\x1...ex))*eu)\n3\n' == 'Frame = (\x1...ex))*eu)\n3\n'
  Skipping 171 identical leading characters in diff, use -v to show
  - eu.ev_r = -1/2 + 1/2
  ?           -    ^
  + eu.ev_r = 1/2 - 1/2
  ?               ^
    ev.eu_r =  0 
    ev.ev_r = 1
    Frame = (#x1B[0;34mex#x1B[0m + #x1B[0;34mey#x1B[0m + #x1B[0;34mez#x1B[0m,#x1B[0;34mex#x1B[0m - #x1B[0;34mey#x1B[0m)
    Reciprocal Frame = (#x1B[0;34mex#x1B[0m/3 + #x1B[0;34mey#x1B[0m/3 + #x1B[0;34mez#x1B[0m/3,#x1B[0;34mex#x1B[0m/2 - #x1B[0;34mey#x1B[0m/2)
    eu.eu_r = 1
  - eu.ev_r = -1/2 + 1/2
  ?           -    ^
  + eu.ev_r = 1/2 - 1/2
  ?               ^
    ev.eu_r =  0 
    ev.ev_r = 1
    eu = #x1B[0;34mex#x1B[0m + #x1B[0;34mey#x1B[0m + #x1B[0;34mez#x1B[0m
    ev = #x1B[0;34mex#x1B[0m - #x1B[0;34mey#x1B[0m
    eu^ev|ex
    (eu^(ev|ex))
    #x1B[0;34mex#x1B[0m + #x1B[0;34mey#x1B[0m + #x1B[0;34mez#x1B[0m
    eu^ev|ex*eu
    ((eu^(ev|ex))*eu)
    3
#x1B[0m

[utensil]

I'll use it for symengine

[eric-wieser]

That was my intent here, yeah

[utensil]

Aha, you use this to make both version happy

[utensil]

We better use the 1.5 output as the default, and replace the old one to the new one here?

[eric-wieser]

We can do that in a follow-up

The sanitizations apply both to the expected and actual result - so will keep working for 1.3.0 CI runs when we regenerate the notebook with 1.5.0

[utensil]

Wonderful!