#60 Unnecessary iFormula-specific options

[zeon_account_will_be_deleted]

What do these settings do in the "Format iFormula" dialog:

1. Inhibit underflow exceptions
2. Evaluate to real roots?

Isn't it pointless? I thought these ones are the document-specific settings. It seems like they don't even change anything when applied for one formula.

I suggest to remove them (if I'm not wrong of course).

[zeon_account_will_be_deleted]

I've found a strange behaviour. I have the equation b=10.0^{-val(a)} and a is a very huge number so b should tend to zero. When I switch the "Inhibit underflow exceptions" option in the iFormula, it is changed to b=10.0^{-a} (val() disappeared! Is it a bug?

[Jan Rheinlaender]

After fixing the bug about the unnecessary options, I can't reproduce the error any more. But also the "inhibit underflow exception" doesn't seem to work at all. Did it ever work before in older versions of iMath?

[Jan Rheinlaender]

That was a bug. The options now work on a line-by-line basis.
Commit 641d093aa4a8fade217eb8bd00bb91a42c8d8781

[Jan Rheinlaender]

I uploaded a new beta version 2.2.5~beta1

[zeon_account_will_be_deleted]

The second example: Unnecessary iFormula-specific options-2.odt.

Now LO crashes when I try to open "Format iFormula" dialog for the second iFormula (see screenshots). Also I deleted the second iFormula and created the same one - the issue repeated, LO crashed.

Evaluate to real root option works correctly for whole document and as iFormula option.

[zeon_account_will_be_deleted]

That's right, "inhibit underflow exception" doen't work at all now. But it did in earlier versions.

When I set the doc-level option "inhibit underflow exception" to false or true the "floating point underflow" error occurs anyway.

If I set "inhibit underflow exception=true" manually in the iFormula the "floating point underflow" error occurs too.

[Anonymous]

I've posted a question on the CLN mailing list about why inhibiting underflow exceptions doesn't work (any more).

[Jan Rheinlaender]

This was a bug in CLN which has been fixed in version >= 1.3.5. Future Windows builds of iMath will use this version, but note that e.g. Ubuntu up to and including 19.10 uses CLN 1.3.4.