418638 - Bugs in JS shell functions scatter() and sleep()

Status: RESOLVED FIXED

(Core :: JavaScript Engine, defect)

Description

[Jason Orendorff [:jorendorff]]

To my shame I posted patch v11 in bug 404879 without testing the new snippet of code.  It's even more busted than v9 and v10, of course.

I also found some other minor bugs, poking around.  scatter() passes PR_LOCAL_THREAD to PR_CreateThread; that should be PR_GLOBAL_THREAD.  sleep() doesn't detect when it's passed 0 arguments.  It doesn't do anything meaningful when an argument that isn't a number.  It should throw in both cases.

Comment 1

[Jason Orendorff [:jorendorff]]

Attachment: v1

On second thought, sleep(NaN) or sleep("string") will do sleep(0).  I would have it throw a RangeError or such, but there's no really good, direct way to do that.  Also JavaScript has a proud tradition of behaving badly in these cases.  Separate bug.

Comment 2

[Jason Orendorff [:jorendorff]]

Attachment: v2

Changes the NaN check to use a macro.

Comment 3

[Brian Crowder]

Comment on attachment 304582 [details] [diff] [review]
v2

Worth speculating about whether JSDOUBLE_IS_NaN should be public; but for another bug.

Comment 4

[Jason Orendorff [:jorendorff]]

Attachment: v3

This additionally fixes a stupid crash when n==0, that is, scatter([]).

Comment 5

[Jason Orendorff [:jorendorff]]

Attachment: v3

How about the right patch.

Comment 6

[Brendan Eich [:brendan]]

As we evolve toward JS2/ES4, ask yourself not what crazy loosy-goosey conversion would JS1 do, but what type annotations would you want on the function you are adding, and what built-in and few conversions might apply.

/be

Comment 7

[Brian Crowder]

js.c:3.200