Move handling of duplicate files

[blueyed]

This removes the hack added in
#3802.

This also improves performance a bit with the test case for #4237:
Before: 7.95s
After: 7.51s

[nicoddemus]

Cool 👍

[blueyed]

One failing test: https://travis-ci.org/pytest-dev/pytest/jobs/446299602#L1014 - maybe just needs adjustment?

[blueyed]

It has an additional <ParsedCall 'pytest_collectstart'(**{'collector': <Package '/tmp/pytest-of-user/pytest-310/test_collect_only_with_various_situations0'>})>, now:

[<ParsedCall 'pytest_collectstart'(**{'collector': <Session 'test_collect_only_with_various_situations0'>})>,
 <ParsedCall 'pytest_collectstart'(**{'collector': <Package '/tmp/pytest-of-user/pytest-310/test_collect_only_with_various_situations0'>})>,
 <ParsedCall 'pytest_collectstart'(**{'collector': <Module 'test_one.py'>})>,
 <ParsedCall 'pytest_collectstart'(**{'collector': <Class 'TestX'>})>,
 <ParsedCall 'pytest_collectstart'(**{'collector': <Instance '()'>})>,
 <ParsedCall 'pytest_collectstart'(**{'collector': <Class 'TestY'>})>,
 <ParsedCall 'pytest_collectstart'(**{'collector': <Instance '()'>})>,
 <ParsedCall 'pytest_collectstart'(**{'collector': <Module 'test_three.py'>})>]

[blueyed]

It was changed to 8 in 2b14108 (Add package scoped fixtures #2283) - which makes sense, but then to 7 again in fedc785 (#3389), likely not intentional?! (ping @jonozzz).

[blueyed]

Went ahead and amended the test, but please ACK.

[jonozzz]

Not sure why there's an extra node in there now.

[blueyed]

@jonozzz
It was changed in fedc785#diff-7eb3d34b4de135c7f717e983c8282210R195 (by you I've assumed - renamed username, or am I wrong)?

[jonozzz]

Went down the blame road, and it's funny that ~8 years ago it started as "== 8" then 3 years ago changed to 7, then 8, 7 and now back to 8. Something's going on in there :)

[blueyed]

Yeah.

I think the initial 8 was due to the "XXX" comment that appears to have been addressed when it changed to 7, and then it went to 8 again when package collection was added.
The question now is why it was changed to 7 again. This might have been accidental, and now that I am removing the HACK here it works as expected again? (i.e. 8)

[codecov]

Codecov Report

Merging #4241 into features will decrease coverage by 0.22%.
The diff coverage is 100%.

@@             Coverage Diff              @@
##           features    #4241      +/-   ##
============================================
- Coverage     95.77%   95.54%   -0.23%     
============================================
  Files           109      109              
  Lines         24351    24348       -3     
  Branches       2404     2402       -2     
============================================
- Hits          23322    23264      -58     
- Misses          733      772      +39     
- Partials        296      312      +16

Flag	Coverage Δ
#docs	?
#doctesting	?
#linting	?
#linux	95.54% <100%> (-0.01%)	⬇️
#nobyte	91.08% <100%> (-0.66%)	⬇️
#numpy	41.32% <42.85%> (-51.55%)	⬇️
#pexpect	41.32% <42.85%> (-0.01%)	⬇️
#py27	93.77% <100%> (-0.12%)	⬇️
#py34	91.96% <100%> (-0.08%)	⬇️
#py35	91.98% <100%> (-0.08%)	⬇️
#py36	93.66% <100%> (-0.19%)	⬇️
#py37	92.16% <100%> (+0.05%)	⬆️
#trial	41.32% <42.85%> (-51.55%)	⬇️
#windows	?
#xdist	93.47% <100%> (-0.13%)	⬇️

Impacted Files	Coverage Δ
src/_pytest/python.py	95.47% <ø> (-0.27%)	⬇️
src/_pytest/main.py	96.22% <100%> (ø)	⬆️
testing/test_session.py	96.45% <100%> (ø)	⬆️
testing/test_paths.py	86.36% <0%> (-13.64%)	⬇️
src/_pytest/assertion/util.py	92.92% <0%> (-5.19%)	⬇️
testing/test_tmpdir.py	94.03% <0%> (-4.64%)	⬇️
src/_pytest/capture.py	90.16% <0%> (-3.21%)	⬇️
src/_pytest/pathlib.py	87.42% <0%> (-2.4%)	⬇️
testing/acceptance_test.py	97.18% <0%> (-1.09%)	⬇️
src/_pytest/nodes.py	93.9% <0%> (-0.82%)	⬇️
... and 4 more

---

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 6dc575f...e041823. Read the comment docs.