Closed lroberts7 closed 1 year ago
Note there aren't any unit tests on this as it's a simple call out to dgl but if there are suggestions around adding a unit test please let me know. Also here is a stacktrace of the issue:
$ dglke_partition -k 1 --data_path '/Users/lroberts/datascience/dgl/dgl-ke/data/codebase'
load dataset..
Reading train triples....
Finished. Read 483142 train triples.
Reading valid triples....
Finished. Read 50000 valid triples.
Reading test triples....
Finished. Read 59071 test triples.
construct graph...
/opt/homebrew/lib/python3.10/site-packages/dgl/_deprecate/graph.py:1023: DGLWarning: multigraph will be deprecated.DGL will treat all graphs as multigraph in the future.
dgl_warning("multigraph will be deprecated." \
partition graph...
Traceback (most recent call last):
File "/opt/homebrew/bin/dglke_partition", line 33, in <module>
sys.exit(load_entry_point('dglke==0.1.0.dev0', 'console_scripts', 'dglke_partition')())
File "/opt/homebrew/lib/python3.10/site-packages/dglke-0.1.0.dev0-py3.10.egg/dglke/partition.py", line 119, in main
AttributeError: module 'dgl' has no attribute 'transform'. Did you mean: 'transforms'?
$ python3 Python 3.10.8 (main, Oct 13 2022, 09:48:40) [Clang 14.0.0 (clang-1400.0.29.102)] on darwin Type "help", "copyright", "credits" or "license" for more information.
import dgl print(dgl.version) 0.9.1
closed via -> https://github.com/awslabs/dgl-ke/pull/270
I noted this in a comment: https://github.com/awslabs/dgl-ke/pull/268/files#diff-6fa7773ef2017429e0385fdbff10e2c9b5021e282a503eb7e47eeb93f651eb18
The version upgrade in
dgl
https://github.com/awslabs/dgl-ke/pull/268 caused thedgl.transform
method in partition.py to raise an error here: https://github.com/awslabs/dgl-ke/blob/276c2e41b0d06bd5881d2873731b4551d0364a41/python/dglke/partition.py#L119 if the environment has a morerecent version of dgl.Dgl version 0.8 changed the name to
transforms
(added ans
) -> https://github.com/dmlc/dgl/releases/tag/0.8.0A fix would be to add an if/else block based on the dgl version and call either attribute. I used the search box in github to search this repository and this appears to be the only use of the
dgl.transform