lib: fix basename comparison on windows

[tflanagan]

Fixes TODO in lib/path.js ref'd here #4642

varf=win32SplitPath(path)[2];// TODO: make this comparison case-insensitive on windows?if(ext&&f.substr(-1*ext.length)===ext){f=f.substr(0,f.length-ext.length);

[mscdex]

This change assumes that ext is lowercase, which may not necessarily be true.

[mscdex]

Also, I'm not sure that the isWindows check is necessary since it's already supposed to be a win32.* method.

[tflanagan]

Thanks @mscdex, I've address those comments

[mscdex]

I think this conditional can be dropped as well, since path.win32 would be available on all platforms and is the same implementations that would be used on real Windows.

[tflanagan]

Thanks @mscdex, I've changed this

[cjihrig]

Would if (ext && f.substr(-1 * ext.length).toLowerCase() === ext.toLowerCase()){ have been enough to address the original comment?

[tflanagan]

It would, but you were concerned with readability in a previous comment, that would work, but would kill readability

[cjihrig]

For readability, it could be: if (ext && f.toLowerCase().endsWith(ext.toLowerCase())){

[tflanagan]

That is much more elegant and witty.

I'll push a fix soon, thanks @cjihrig.

[tflanagan]

/poke

Would this be a semver-minor change?

[bnoordhuis]

It's a backwards incompatible change so semver-major, I'll add the tag.

Apropos .toLowerCase(), it has a possible gotcha in that it may not correctly lower-case some Unicode characters.

It's currently implemented using the unibrow library but I know that the V8 team has plans to swap it out for ICU because unibrow lags behind the Unicode spec. To wit, V8's test suite has a number of regression tests for .toLowerCase() that are marked FAIL.

In other words, this may not be an entirely risk-free change.

[tflanagan]

/poke

[jasnell]

/cc @bnoordhuis@mscdex@orangemocha ... any thoughts on this one?

[orangemocha]

No objections to doing case-insensitive matching, but this function has undergone substantial changes (and the todo was removed) with b212be0. Could you please rework your changes over the current master branch?

[jasnell]

@tflanagan .. ping

[MylesBorins]

@tflanagan ping

[Trott]

@tflanagan Same question here: Any chance you can do a rebase? If not, I or someone else can try to take it on, but it would be awesome to get this landed and it would be great to see you back in action. (But I totally get it if you're busy with other stuff or your interests have gone elsewhere!)

[jasnell]

Closing given the lack of forward progress on this.

[gibfahn]

For posterity, I think this was fixed as part of @mscdex's rewrite in #5123

b212be0#diff-c0fae59301ef49fb71b43278db9ad881L344