Backport #5272: Float#round doesn't round big values

• Status: Closed
• Assignee: -

[marcandre (Marc-Andre Lafortune)]

For large enough values (outside of Fixnum range), Float#round does not round at all but simply truncates:

2.999999999999999e20.round(-20) # => 200000000000000000000

Fixed in trunk, but would be nice to backport for Ruby 1.9.3.

[marcandre (Marc-Andre Lafortune)]

• Status changed from Open to Closed
• % Done changed from 0 to 100

This issue was solved with changeset r33186.
Marc-Andre, thank you for reporting this issue.
Your contribution to Ruby is greatly appreciated.
May Ruby be with you.

---

• numeric.c (flo_round): Make Float#round round big values [bug #5272]

[marcandre (Marc-Andre Lafortune)]

• Status changed from Closed to Open

Opening the backport request. Note: dependent on r33185 and r33183

[nagachika (Tomoyuki Chikanaga)]

I've found that TestFloat#test_round fails after r33186 on i686-linux.

For (4294967295.8).round, 32bit version ruby returns 4294967295, while 64bit version(x86_64-linux) returns correctly 4294967296. It seems that when result value is Bignum Float#round works like floor().

[nagachika (Tomoyuki Chikanaga)]

• File numeric.c.patch numeric.c.patch added

Attach a tiny patch. It works for me with make test-all.

[marcandre (Marc-Andre Lafortune)]

Tomoyuki Chikanaga wrote:

Attach a tiny patch. It works for me with make test-all.

Thanks! You patch works, but the bug was actually in dbl2ival. See r33199 / #5276

[naruse (Yui NARUSE)]

• Tracker changed from Backport to Bug
• Status changed from Open to Assigned
• Assignee set to marcandre (Marc-Andre Lafortune)

[naruse (Yui NARUSE)]

• ruby -v set to ruby 1.9.4dev (2011-08-05 trunk 32856) [x86_64-darwin11.0.0]

This issue still happends:
Float("2.5e-22").round(22) #=> 2.0e-22aaaaaa

[naruse (Yui NARUSE)]

• ruby -v changed from ruby 1.9.4dev (2011-08-05 trunk 32856) [x86_64-darwin11.0.0] to ruby 1.9.4dev (2011-09-06 trunk 33199) [x86_64-darwin11.1.0]

[marcandre (Marc-Andre Lafortune)]

• Category set to core
• Status changed from Assigned to Open
• Assignee deleted (marcandre (Marc-Andre Lafortune))

Yui NARUSE wrote:

This issue still happends:
Float("2.5e-22").round(22) #=> 2.0e-22

This is a different issue, caused by calculation error. I opened issue #5273 to deal with boundary cases like this one. This present issue was not due to approximation but by truncation of any values.

Moving back to backport request

[marcandre (Marc-Andre Lafortune)]

• Tracker changed from Bug to Backport

[kosaki (Motohiro KOSAKI)]

If I parsed this thread correctly, this bug is not a regression. therefore it isn't 1.9.3p0 material. I think.

[marcandre (Marc-Andre Lafortune)]

See my comment on #5276

[naruse (Yui NARUSE)]

• Target version deleted (1.9.3)

[naruse (Yui NARUSE)]

• Status changed from Open to Closed

This issue was solved with changeset r33898.
Marc-Andre, thank you for reporting this issue.
Your contribution to Ruby is greatly appreciated.
May Ruby be with you.

---

merge revision(s) 33158,33185,33186:

* numeric.c (flo_round): substitute machine dependent magic number.

* numeric.c (flo_round): Make Float#round round big values [bug
  #5272]