Why source deletion is shown for this js Source file ? about gumtree

[jrfaller]

Can you attach the two files ?

[irajdeep]

Sure
left.js.zip
right.js.zip

[jrfaller]

Ok same result here,

What's going on is that there are many little differences so GumTree do not recognize the function definition as being the same, therefore evertyhing seem to be moved even if it's not really the case.

There are some options that can make it better, with the following config it's way more accurate (pass to VM options): -Dgumtree.match.bu.sim=0.7

Of course it should be documented somewhere ;)

[irajdeep]

Can I pass this to the gumtree command , something like
./gumtree webdiff left.js right.js -Dgumtree.match.bu.sim=0.7 (because it's not working like this :) ) or should I add it somewhere else ?

[jrfaller]

Good question, I will ask it to floreal.

[irajdeep]

Sure

[morandat]

I have no clue of who did this script (maybe gradle) but it has no chance to work like this.
You can try (let me know which one are working):
./gumtree -Dgumtree.match.bu.sim=0.7 webdiff left.js right.js
JAVA_OPTS="-Dgumtree.match.bu.sim=0.7" ./gumtree webdiff left.js right.js
or (but I'm not sure we have implemented it)
./gumtree "-c match.bu.sim=0.7" webdiff left.js right.js

If none work I'll try it tomorow

[irajdeep]

The second command
JAVA_OPTS="-Dgumtree.match.bu.sim=0.7" ./gumtree webdiff left.js right.js worked.
Thanks

[irajdeep]

umm , at instances I am finding the JAVA_OPTS="-Dgumtree.match.bu.sim=0.7" flag is showing more differences than there actually are in the source files . Can someone explain why it might be ?

[jrfaller]

It's not a flag it's a threshold. If you want to know more, you can read our ASE paper where it is described.