1615600 - BigInt(fractionalNumber) results in misleading error message

Status: RESOLVED FIXED

(Core :: JavaScript Engine, defect)

Description

[Michał Wadas]

User Agent: Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:73.0) Gecko/20100101 Firefox/73.0

Steps to reproduce:

Typed in console:

BigInt(2.5)

Actual results:

RangeError: can't convert non-finite number to BigInt

Expected results:

RangeError: can't convert non-integer number to BigInt

Comment 1

[Virginia (please send need info request to Brindusa Tot)]

Hi michalwadas,

Thank you for your report.

I was able to reproduce it on my end on Ubuntu 18.04 (64-bit) using:

• Firefox Nightly 75.a01 (2020-02-18),
• Firefox Beta 74.0b4
• Firefox Release 71.0, 72.0.2 and 73.0.
• Firefox 68.5.0esr

I will add this issue to the DevTools - Console component so their team can take a look and give their insights on this bug.

Regards,
Virginia

Comment 2

[Nicolas Chevobbe [:nchevobbe]]

Thanks for the report.
For what it's worth Chrome is displaying:

Uncaught RangeError: The number 2.3 cannot be converted to a BigInt because it is not an integer

The error we display is defined in js/src/js.msg#721 and called from js/src/vm/BigIntType.cpp#1784-1788
We can see we're only checking if the number is an integer.

Let's ask Andy and Jeff what they think of this.

Comment 3

[Andy Wingo [:wingo]]

To me this is indeed a bug! I think we should just change the text of the message, probably to be closer to Chrome (i.e. ideally including the value).

Comment 4

[Jeff Walden [:Waldo]]

Attachment: Bug 1615600 - Use a better error message for when an attempt is made to convert a non-integer number to a BigInt. r=wingo

Comment 5

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/499d0e825c87
Use a better error message for when an attempt is made to convert a non-integer number to a BigInt.  r=wingo

Comment 6

[Alexandru Michis [:malexandru]]

https://hg.mozilla.org/mozilla-central/rev/499d0e825c87

Comment 7

[Pascal Chevrel:pascalc (PTO until Sept 2)]

Is that a patch we should uplift to beta or can it ride the 75 train?

Comment 8

[Jeff Walden [:Waldo]]

Subpar and even misleading error messages are not something we should be uplifting. The actual language semantics are correct (error message contents are not specified), which is what matters most. (And even if they weren't, this is fairly corner-case and would not affect most properly-written code.)