[Filing bugs] Error：scikit-learn/sklearn/utils/fixes.py

[willduan]

There is a new bug in master brunch, scikit-learn/sklearn/utils/fixes.py
in line 406: if np_version < (1, 12, 0): , an error occured, TypeError: unorderable types: str() < int().
It's because in the function :

def _parse_version(version_string):
    version = []
    for x in version_string.split('.'):
        try:
            version.append(int(x))
        except ValueError:
            # x may be of the form dev-1ea1592
            version.append(x)
    return tuple(version)

However， my numpy version is：1.12.0b1, this funciotn return a tuple(1,12,'0b1') , but in line 406, It is compared with integer type, so the error occured.
I just in a simple way to solve it. I change the code to if np_version < (1, 12, '0'): in line 406, just change the third tuple index to string type.
Of course, this is not a common method.

[jnothman]

I think np_version[:2] < (1, 12) would be more robust. Please submit a
PR, and perhaps identify other cases of comparison in fixes.py where we are
at risk of breaking for non-releases.

On 17 November 2016 at 14:09, willduan notifications@github.com wrote:

There is a new bug in master brunch, scikit-learn/sklearn/utils/fixes.py
in line 406: if np_version < (1, 12, 0): , an error occured, TypeError:
unorderable types: str() < int().
It's because in the function :
def _parse_version(version_string): version = [] for x in
version_string.split('.'): try: version.append(int(x)) except ValueError: #
x may be of the form dev-1ea1592 version.append(x) return tuple(version)
However， my numpy version is：1.12.0b1, this funciotn return a
tuple(1,12,'0b1') , but in line 406, It is compared with integer type, so
the error occured.
I just in a simple way to solve it. I change the code to if np_version <
(1, 12, '0'): in line 406, just change the third tuple index to string
type.
Of course, this is not a common method.

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
#7898, or mute the
thread
https://github.com/notifications/unsubscribe-auth/AAEz6ylNZpJrZ3d8QXWO_zb_n6NuPZdEks5q-8VYgaJpZM4K0yOg
.

[amueller]

Can't we use distutils.LooseVersion?

[lesteve]

Can't we use distutils.LooseVersion?

Just to prevent duplicating work, the short answer is no and the longer answer is in the PR #7902.

[Cli98]

Clearly slight modifications on parse function could solve the problem. Take the non-number characters out may be enough.
we can also type version number string directly to the fields to past the test.

[Maneetgiri]

I am not able to open the Fixes.py in my system. I tried opening with 'Edit with IDLE 3.5' but it doesn't open any window. is there anything else that can be done to open and edit that particular file.

[nelson-liu]

@Maneetgiri try opening it with just a regular text editor (e.g. notepad, sublime text, emacs, vim, etc).

[biotechie]

Thanks, willduan. The fix worked for me.

[jliu118]

@jnothman The ticket is closed but the issue still there:
when you add "x" into list "version"(change to tuple) at return statement if x is NOT digits, this will cause type error later when you compare tuple (1,12,'1rc1) to tuple (1,12,0). Two possible solution.

1. Use tuple of string.
   Change version.append(int(x))
   to version.append(x)
   Comparison using tuple of strings.
   Change tuple (1,12,0) to ('1', '12', '0') etc.
   A lot of changes.
2. Simple fix. Extract numeric portion of x if x contains letters using regular expression. Comparison keeps the same.

Hope somebody will merge my fix into next release because I'm NOT a contributor.
...
import re
...
def _parse_version(version_string):
version = []
for x in version_string.split('.'):
try:
version.append(int(x))
except ValueError:
# x may be of the form dev-1ea1592
# version.append(x)
digits=re.match(r'^(\d)+\D',x)
if digits:
version.append(int(digits.group(1)))
else:
# use 0 if no leading digits found
version.append(0)
return tuple(version)

Test environment:
Python 3.4
Numpy 1.12.1rc1
sklearn 0.18.1

[lesteve]

Look at #7980 which is a more generic issue about version comparison across our codebase. There is already a PR associated to this issue by the way.

[lesteve]

Look at #7980 which is a more generic issue about version comparison across our codebase. There is already a PR associated to this issue by the way.

And actually I believe your particular issue is already fixed in master, i.e. the development version.