UnboundLocalError: local variable 'labels' referenced before assignment

[StephaneLu]

UnboundLocalError: local variable 'labels' referenced before assignment

[GGiecold-zz]

Hi,

Could you please share more information regarding the circumstances
surrounding this issue?

Are you running Concurrent_AP from the command line, by calling
'Concurrent_AP file_path', with file_path referring to a tab-separated
dataset with each row standing for a sample?

Just this morning, I ran this module both on a laptop running Fedora-Linux
and on my MacBook; I could not reproduce the error message you're reporting.

Kind regards,

Gregory
On Mar 11, 2016 2:22 AM, "StephaneLu" [email protected] wrote:

Hi,
I am having an issue running your code on mac osx :

Traceback (most recent call last):
File "/usr/local/bin/concurrent_AP", line 9, in
load_entry_point('Concurrent-AP==1.4', 'console_scripts',
'Concurrent_AP')()
File "/Library/Python/2.7/site-packages/Concurrent_AP.py", line 1416, in
main
output_clusters(labels, cluster_centers_indices)
UnboundLocalError: local variable 'labels' referenced before assignment

Any hint ?

—
Reply to this email directly or view it on GitHub
#2.

[StephaneLu]

Hi Gregory,
thank you for your work 👍 ,
I fixed the issue by changing the np.savetxt() with the proper delimiter. '\t'
I was trying to load the hdf5 file prior to create it ...
I disabled the fh = NamedTemporaryFile('w', delete = False, dir = './',suffix = '.h5') to ba able to load it in the future.
I made several other changes on memory because osx does not support '/proc/meminfo' :
memory() : psutil.phymem_usage().free / used, it seems to work so far ...
get_chunksize() : as max(xxxxx, mem_free * 0.07)

Regards
Stéphane

[GGiecold-zz]

Hello Stéphane,

Glad to hear everything is in order!

Your message made me realize that the documentation to Concurrent_AP didn't
emphasize enough that it only accepts tab-separated data files. I've edited
its PyPI page accordingly (the Github page had been edited separately and
was clearer in this regard). Thanks!

I'm swamped with work right now but I should change at some point the code
relying on /proc/meminfo. Some one trying to run DBSCAN_MULTIPLEX on
Windows had a similar problem a while ago.
Feel free to push the change you brought to get things working on OS X.

All the best with your programming endeavors,

Gregory
On Mar 11, 2016 11:48 AM, "StephaneLu" [email protected] wrote:

Hi Gregory,
thank you for your work [image: 👍] ,
I fixed the issue by changing the np.savetxt() with the proper delimiter.
'\t'
I was trying to load the hdf5 file prior to create it ...
I disabled the fh = NamedTemporaryFile('w', delete = False, dir =
'./',suffix = '.h5') to ba able to load it in the future.
I made several other changes on memory because osx does not support
'/proc/meminfo' :
memory() : psutil.phymem_usage().free / used, it seems to work so far ...
get_chunksize() : as max(xxxxx, mem_free * 0.07)

Regards
Stéphane

—
Reply to this email directly or view it on GitHub
#2 (comment)
.

[lucapinello]

Hi @StephaneLu do you mind to share the changes? I have the same problems on my mac.

Thanks!