bcmath: strip leading/trailing zeros option (code is already there)

[quakemmo]

Description

@SakiTakamachi It seems that you're the one taking care of bcmath at least recently.

Would it be possible to add a flag to the bc* functions such as bcadd, bcmul and so on, that would strip the trailing and/or leading zeros from the results? Maybe even make it the default? At this time, bscale(10); bcadd('1', '2'); will produce the result '3.0000000000', which while correct, isn't really ideal.

Or at the very least, give public access to the already existing functions bc_rm_trailing_zeros() and _bc_rm_leading_zeros() ?

Ideally it would also have a flag to either keep or not the integer zero in values such as 0.1234 (so the strip leading zeros would either keep it or just return .1234).

I'm using bcmath very extensively in a project where performance matters and doing such sanitization in PHP itself is 1) a pain and 2) a performance sink. It would really help if that's something that could be handled directly by bcmath, which is otherwise a wondeful library that thankfully exists.

Thank you for your work on bcmath.

[SakiTakamachi]

Thank you!

Since BCMath uses arbitrary precision, making that behavior the default is probably not a very good idea.

If we add too many arguments, things will get complicated, so in this case, I think it would either take an additional argument in the form of a bitmask, or an enum specifying the mode for handling zero.

However, in any case, this would mean making a significant change to the API, so an RFC would be necessary for that.

If you’re interested in pursuing this, I recommend discussing it on the internal mailing list.