Bad isdir result

[agroce]

import shutil
import pyfakefs.fake_filesystem
import os

ROOT_PATH = "/Volumes/ramdisk/test"

fs = pyfakefs.fake_filesystem.FakeFilesystem()
fs.CreateDirectory('/Volumes')
fs.CreateDirectory('/Volumes/ramdisk')
fs.CreateDirectory('/Volumes/ramdisk/test')
os0 = pyfakefs.fake_filesystem.FakeOsModule(fs)
opener0 = pyfakefs.fake_filesystem.FakeFileOpen(fs)

if False:
    os0 = os
    shutil.rmtree(ROOT_PATH)
    os.mkdir(ROOT_PATH)

path0 = ROOT_PATH
component0 = "" 
path0 += os.path.sep + component0 
result = os0.path.isdir(path0) 

print result

[agroce]

(Note that fixing this one will help testing a good bit; incorrect isdir turns out to cause some other issues, when it tries to avoid doing certain operations on directories; I have a big pile of issues from a new test run, but think several are this problem)

[mrbean-bremen]

Yes, this one is MacOS-specific (in the sense that the bug is a regression that only appears under MacOS due to another fix that handled MacOS-specific behavior). I think I have a fix, but I have to wait for the Travis MacOS build to be sure - they currently have problems with MacOS builds (again).
Generally, the rate of regressions here is way too high - this is due to the fact that the behavior for these cases (trailing separator stuff, but also buffering stuff with open/seek/tell) is quite erratic (at least it seems so to me), and usually differs between the different OSes, so what I'm doing here is basically TDD without really understanding the problems.

[agroce]

Yeah, the logic here is sort of random nonsense for each OS, since it's not POSIX-specified. We could drop trailing separators for now, and look for more interesting issues, though some of these can cause some weird behaviors if you assume one OS then migrate to pyfakefs, I guess.

[agroce]

(I had the same Travis mac issue, so just do TSTL mac tests on the laptop I develop on anyway)