-
Notifications
You must be signed in to change notification settings - Fork 13
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Skip failing windows unittests, and omit max_rss on windows (+ correctly report for OS X) #166
Conversation
This is mostly there now I think, but still fails on Windows AppVeyor images with "lmdb error There is not enough space on the disk.". This google post says "On Windows the OS sets the filesize equal to the mapsize", and tensorpack/tensorpack#1209 (comment) recommends "starting with a small map_size on windows but increasing gradually when the dataset is full". Not sure if this helps |
Great stuff, thanks @hyanwong! I agree this is good enough for now - how about we put int a skipIf(windows) like here for the failing tests and open an issue to track the ones that look fixable? Re psutil, we could just leave this out - it's a minor feature that windows users can live without. Or, if psutil is a light and easy dependency, we just go with this for everything. |
Yes, that sounds a quick work-around. I'll check if it works on Windows with enough disk space, then add the hack with a comment.
Annoyingly, although psutil makes |
Makes sense, thanks. I would drop it personally - I can't see anyone doing really large scale stuff on windows, which is where kind of reporting is useful. |
Agree. Will cut it out. |
7af100b
to
645a79e
Compare
Codecov Report
@@ Coverage Diff @@
## master #166 +/- ##
==========================================
- Coverage 93.94% 93.72% -0.22%
==========================================
Files 11 11
Lines 2874 2885 +11
Branches 504 507 +3
==========================================
+ Hits 2700 2704 +4
- Misses 144 148 +4
- Partials 30 33 +3
Continue to review full report at Codecov.
|
645a79e
to
bdfd18d
Compare
LMDB/lmdb@fb5a768 implies that it should be possibly to dynamically grow the lmdb database on windows, so I don't know why it's a problem. Investigating further. |
So it looks like the patch to make this work on windows simply hasn't made it into LMDB 0.9.22 (or even 0.9.23). See mozilla/lmdb-rs#23 and mozilla/lmdb-rs#40 Looks complicated! I don't know if there's a way to fix this, other than by having some sort of conditional code which decreases |
Let's just try reducing the map size for now (on windows) to get CI working, and skipIf any tests that don't behave. There's still value in getting this mostly working on Windows, we can figure out a good way of dealing with it later. |
20a515a
to
3e1f87d
Compare
I set the
and
There's also this:
|
3e1f87d
to
384454f
Compare
Just skipIf these - it's not worth worrying about them. Otherwise looks good! |
257f091
to
d716a2e
Compare
…tly report for OS X) Also add in python-lmdb to get windows compilation working
7c120f6
to
c04a6c9
Compare
Yay. This now passes @jeromekelleher. I've rebased. Not sure how to deal with the coverage though, as it's only windows-specific - do you want a test to see if the max_rss is correctly reported in Unix vs absent in win32? |
Thanks @hyanwong, this is fantastic! Let's not worry about the codecov report. Merging! |
Also add in other packages to get windows compilation working