incorrect RoundUp conversion of large strings

[stevengj]

This is wrong, since the result should be exactly representable in Dec128:

julia> Dec128(big(10)^1000, RoundUp)
1.000000000000000000000000000000001e1000

Unfortunately, this might be a bug in the underlying library? All we are doing is converting the BigInt to a string and then calling the library's from_string routine.

As a result, we get incorrect results like:

julia> d128"1e100" == big(10)^100 # correct
true

julia> d128"1e1000" == big(10)^1000 # incorrect
false

[jmkuhn]

The results vary run to run. Most likely a buffer overflow in the C. I'll look into it.

julia> Dec128("1" * "0"^106, RoundUp)
1.000000000000000000000000000000001e106

julia> Dec128("1" * "0"^105, RoundUp)
1.0e105

julia> exit()
CD0004431999 1006% julia
               _
   _       _ _(_)_     |  Documentation: https://docs.julialang.org
  (_)     | (_) (_)    |
   _ _   _| |_  __ _   |  Type "?" for help, "]?" for Pkg help.
  | | | | | | |/ _` |  |
  | | |_| | | | (_| |  |  Version 1.9.4 (2023-11-14)
 _/ |\__'_|_|_|\__'_|  |  
|__/                   |

julia> using DecFP

julia> Dec128("1" * "0"^106, RoundUp)
1.0e106

[jmkuhn]

I have a patch for this one. I want to do a bit of testing on aarch64, musl, and perhaps riscv64 before releasing a new DecFP_jll.

[jmkuhn]

While testing a patch for this I discovered one additional issue

julia> Dec64("9.99999999999999900000005", RoundUp)
1.0e-7

This issue is also fixed by #194 and JuliaPackaging/Yggdrasil#12397