Fix Windows support for upwards recursive .jsbeautifyrc fallback (originally #37)

[BlaM]

Not sure what I'm doing wrong, but all this package seems to do on the two Windows machines I tried it on is to lock up Atom.

As soon as I press ALT+CTRL+B the whole editor window freezes and after some time a window pops up telling me that something got stuck and asks me if I want to continue waiting or if I want to halt it. Waiting seems to go on forever.

Atom v0.125.0
Windows 7 64bit - one with English locals, one with German locals.

Tried with HTML and JS, both with the same result.

[Glavin001]

Can you also confirm what version of Atom Beautify you are running?

[BlaM]

I tried with versions 0.7.0 (via install last week) and 0.10.0 (via update today).

[BlaM]

Not sure if it's noteworthy: I installed them through the "Packages" interface in Settings, not via command-line "apm".

[Glavin001]

And 0.10.0 same issue? Strange. I have not seen this before. I will try 0.10.0 and a few different laptops I have and see if I can reproduce. Thanks!

[BlaM]

Just to be sure I tried a clean reinstall (completely remove package before installing it again), but with the same result: Locking up - and yes, also with 0.10.0.

I attempted to reformat a 40 line JS file.

Is there anything I can do to gather more info to help you?

[Glavin001]

Including the 40 line JS file contents here so I could test with it would be ideal, however that is up to you if you're in a position to share that source code.

[BlaM]

The JS file does not make a difference. It also happens with any other file. I just confirmed with this one:
https://github.com/h5bp/html5-boilerplate/blob/master/src/js/plugins.js

[Glavin001]

Could you check your JavaScript console in Developer Tools?

[BlaM]

I can open it and I can find beautify.coffee in there. I can have a look at whatever you like if you can give me specific instructions - but so far I don't have the time (or - to be honest - the motivation) to learn CoffeeScript.

[Glavin001]

I want you to just check for errors in the log.

[BlaM]

Okay, I singlestepped through it for a while and now it seems to be stuck in a bit of code here:
https://github.com/Glavin001/atom-beautify/blob/master/lib/beautify.coffee#L252

Not absolutely sure what it does, but my guess is that it starts at some directory and then traverses up until it reaches "/" (root). However in my case (Windows) that never happens, because my root currently is "T:" (that's the content of that variable p).

[RangerMauve]

I've also had this issue. But I only noticed it when I updated to 0.10.0. (Windows 8)

[Glavin001]

Oh, of course! That absolutely makes sense. Thank you for diagnosing this issue @BlaM. I don't have a Windows machine to test on so I neglected to think of that case. Will get right on it.

[Glavin001]

Could you change:

while p isnt "/"

to:

while p isnt path.resolve(p,"../")

And let me know if that works.

[RangerMauve]

It gets up to C:\ and all subsequent calls to path.resolve return the same value.

Maybe count the number of path.sep in the string and return when it's only one?

Probably with this:

while p.split(path.sep).length isnt 2

[Glavin001]

It gets up to C:\ and all subsequent calls to path.resolve return the same value.

Exactly.

So,

while p isnt path.resolve(p,"../")

should equate to

while "C:\" isnt "C:\"

which of course will fail and stop the while loop.

Did it not work?

[RangerMauve]

Sorry, I made a mistake recreating a minimal example. It works just as expected.

[Glavin001]

Okay, if that is confirmed working then I will publish a patch.

[Glavin001]

Published to v0.10.2

Thanks for your patient and support everyone!