0.4.36 cbBuild fails due to egg file accessed through open method #72
Comments
|
I wonder why are md5 being checked for files that come with the installation... |
|
Well, it's not recommended, you're right. It's similar to Python file
handles that should be closed but in practice don't need to be closed. :-)
This worked until now for you and it works on every platform that I tested.
How come that it suddenly fails...? Did you change something on your
machine?
…On Fri, Feb 8, 2019 at 10:43 AM Pablo Moreno ***@***.***> wrote:
Running cbBuild on a config file generated through some of the available R
code here generates the following error:
INFO:root:Found 1 datasets
Traceback (most recent call last):
File ***@***.***/bin/cbBuild", line 11, in <module>
load_entry_point('cellbrowser==0.4.36', 'console_scripts', 'cbBuild')()
File ***@***.***/lib/python3.7/site-packages/cellbrowser-0.4.36-py3.7.egg/cellbrowser/cellbrowser.py", line 3069, in cbBuildCli
File ***@***.***/lib/python3.7/site-packages/cellbrowser-0.4.36-py3.7.egg/cellbrowser/cellbrowser.py", line 2916, in cbBuild
File ***@***.***/lib/python3.7/site-packages/cellbrowser-0.4.36-py3.7.egg/cellbrowser/cellbrowser.py", line 2946, in build
File ***@***.***/lib/python3.7/site-packages/cellbrowser-0.4.36-py3.7.egg/cellbrowser/cellbrowser.py", line 3272, in cbMake
File ***@***.***/lib/python3.7/site-packages/cellbrowser-0.4.36-py3.7.egg/cellbrowser/cellbrowser.py", line 3226, in makeIndexHtml
File ***@***.***/lib/python3.7/site-packages/cellbrowser-0.4.36-py3.7.egg/cellbrowser/cellbrowser.py", line 3165, in writeVersionedLink
File ***@***.***/lib/python3.7/site-packages/cellbrowser-0.4.36-py3.7.egg/cellbrowser/cellbrowser.py", line 2499, in md5WithPython
NotADirectoryError: [Errno 20] Not a directory: ***@***.***/lib/python3.7/site-packages/cellbrowser-0.4.36-py3.7.egg/cellbrowser/cbWeb/ext/jquery-ui-1.12.1.css'
My understanding is that you cannot access a file within an egg through
open, as it is done here:
https:/maximilianh/cellBrowser/blob/f1b0d067f53fe6dee8efec9388582cea8ff4eab1/src/cbPyLib/cellbrowser/cellbrowser.py#L2524
with file
lib/python3.7/site-packages/cellbrowser-0.4.36-py3.7.egg/cellbrowser/cbWeb/ext/jquery-ui-1.12.1.css
inside the egg.
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
<#72>, or mute the thread
<https:/notifications/unsubscribe-auth/AAS-TWTQiGsB247SGFmAqzx28NaL03uTks5vLcU5gaJpZM4axB09>
.
|
|
Well, I have to say I haven't got this far before with this newest versions, and maybe I'm one of the few using cellbrowser through a conda installation. But I'm always doing essentially the same: this is on a fresh installation of 0.4.36 using conda on a fresh conda environment (I haven't pushed the 0.4.36 changes to bioconda since it seems to be broken on a canonical execution). Do your other installation methods also produce eggs? or do they produce wheels (maybe only the conda installation produces eggs? or maybe this only affects distribution?) |
|
I have left here https://drive.google.com/file/d/1qRBnYejfcepFrQa3eotLPx7AljQmANnd/view?usp=sharing a tar.gz with the inputs for Checking out https:/ebi-gene-expression-group/bioconda-recipes/tree/cellbrowser-0.4.36 and doing: should reproduce the issue (hopefully). I had to move away from using cbImportSeurat2, as the conda package adding seurat as a dependency was very large and took a while to install everytime. So now I create the files for cellbrowser inside a seurat container using the R script you have in this repo taken from seurat3. |
|
I will try and see whether the latest tagged version exhibits the same problem, but judging by the code I presume is probably the case still. |
|
An interesting discussion here: https://stackoverflow.com/questions/49782111/access-a-file-from-a-python-egg |
|
Yes, you're absolutely right. I have to fix this.
I am still curious why this has never happened to me or anyone else I
know....
Did you change something in the way you install these packages? Did you run
a normal pip install to get the package installed?
…On Fri, Feb 8, 2019 at 1:55 PM Pablo Moreno ***@***.***> wrote:
An interesting discussion here:
https://stackoverflow.com/questions/49782111/access-a-file-from-a-python-egg
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#72 (comment)>,
or mute the thread
<https:/notifications/unsubscribe-auth/AAS-TY-kATpmN8dpQl9PPwcBaCqznRScks5vLfI2gaJpZM4axB09>
.
|
|
No, I don't think that conda uses pip to install python packages, but rather use setuptools or alike directly, which probably leaves the package installed in a slightly different manner. |
|
I can confirm that the error doesn't show up when installing through pip, as pip doesn't produce eggs and leaves the directory structure complete and not inside the zipped egg. |
|
The egg format is not supported anymore and the community has moved away from it. Pip removed support for it pypa/pip#1749 To make sure that conda doesn't try an egg install anymore (how did you obtain the egg? I'm unable to get one, no matter what I specify) I've added an option now to setup.py to make sure that it always creates a wheel. Thanks! |
|
No worries, I switched the way that the conda package is installed, so it doesn't use eggs anymore. |
Running cbBuild on a config file generated through some of the available R code here generates the following error:
My understanding is that you cannot access a file within an egg through open, as it is done here:
cellBrowser/src/cbPyLib/cellbrowser/cellbrowser.py
Line 2524 in f1b0d06
with file
lib/python3.7/site-packages/cellbrowser-0.4.36-py3.7.egg/cellbrowser/cbWeb/ext/jquery-ui-1.12.1.cssinside the egg.The text was updated successfully, but these errors were encountered: