Documentation request for arb_get_str()

[Martin Fuller]

I'd like to request an extra line of documentation in arb_get_str():

"The output radius may be massively larger than the stored radius if the output midpoint is rounded. If this is not what you want, consider printing the radius separately using  arb_radref()."

This line would have saved me an hour or more and I think it would have prevented https://github.com/flintlib/flint/issues/2208 .

[Joel Dahne]

This has been a common source of opened issues in the Julia wrapper
Arblib.jl as well. To the extent that we have added extra documentation
to describe this
https://kalmarek.github.io/Arblib.jl/stable/printing/#Printing

Best,
Joel

[Martin Fuller]

Thank you for posting the link. I found the second paragraph really helpful - "when the radius is too large for even the first digit to be known ..." - because it goes on to describe a concrete example and some of the implications of the output convention. Please can an example like this be added to the 3rd line of the documentation for arb_get_str()?

My problem was more like the third paragraph but much larger. My data was something like [1.5...e+15 +/- 1e-9]. The default output seems to have 6 digits, which meant the radius was output as 1e+9 instead of 1e-9. It would be nice if the documentation mentioned this feature and pointed out that other options are available.

PS is there any way to access more=True or variable digits via flint_printf? If not, please add a line into https://flintlib.org/doc/flint.html#input-output to say that all the objects have their own _print functions which offer more control.