Disable the minimum number of matches and common code thresholds by default #45
Conversation
There is no point to setting min-matches to a value above 0 by default. Even though it may yield a few project pairs that have very few matches, it may also eliminate project pairs with large, valid matches after match expansion. A similar argument could be made for disabling `common-code-threshold` by default: For a small set of submissions (< 20), the threshold corresponds to < 2 submissions which implies no match gets reported.
|
Good point! I think it would be a good idea to also set the common code threshold to zero, as you suggested. It might be nice to add a "verbose" flag which makes the tool print out various stats like number of projects found, number of project pairs, number of project pairs after applying min-matches, etc. That way users can figure out which command-line arguments are problematic. |
Yes we should move that to another issue though. |
Gadiguibou
force-pushed
the
disable-min-matches-by-default
branch
from
0f43ef5 to
fe06dd3
Compare
Gadiguibou
changed the title
min-matches to 0
Gadiguibou
changed the title
I created a new issue for it: #46. |
There is no point to setting
min-matchesto a value above 0 by default. Even though it may yield a few project pairs that have very few matches, it may also eliminate project pairs with large, valid matches after match expansion.A similar argument could be made for disabling
common-code-thresholdby default: For a small set of submissions (< 20), the threshold corresponds to < 2 submissions which implies no match gets reported.