expm1 and log1p

[kbriggs]

BPO	1163139
Nosy	@tim-one, @rhettinger

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = None
closed_at = <Date 2005-06-03.01:34:52.000>
created_at = <Date 2005-03-14.17:24:46.000>
labels = ['type-feature', 'library']
title = 'expm1 and log1p'
updated_at = <Date 2005-06-03.01:34:52.000>
user = 'https://bugs.python.org/kbriggs'

bugs.python.org fields:

activity = <Date 2005-06-03.01:34:52.000>
actor = 'rhettinger'
assignee = 'none'
closed = True
closed_date = None
closer = None
components = ['Library (Lib)']
creation = <Date 2005-03-14.17:24:46.000>
creator = 'kbriggs'
dependencies = []
files = []
hgrepos = []
issue_num = 1163139
keywords = []
message_count = 5.0
messages = ['54414', '54415', '54416', '54417', '54418']
nosy_count = 3.0
nosy_names = ['tim.peters', 'rhettinger', 'kbriggs']
pr_nums = []
priority = 'normal'
resolution = 'rejected'
stage = None
status = 'closed'
superseder = None
type = 'enhancement'
url = 'https://bugs.python.org/issue1163139'
versions = []

[kbriggs]

Could we please have expm1 and log1p interfaced in the
math module (where these exist in libm)? These are
essential for any serious mathematical applications.
lgamma would be nice too.

[tim-one]

Logged In: YES
user_id=31435

Hard one, since none of those functions are required by the
C89 standard, Python doesn't require more than C89, and the
math module overwhelmingly consists of thin wrappers
around the platform C's implementations. When C99 is
universally available, then it will be easy. Before then,
someone would have to volunteer non-trivial work (for
example, Microsoft's C runtime doesn't supply any of these --
it's a x-platform mess).

[kbriggs]

Logged In: YES
user_id=888261

Is there an objection in principle to including something in the
math module only if it is found in the system's libm?

If so, a solution would be to use a C implementation of expm1
when it is not already supplied, for example from fdlibm:
http://www.netlib.org/fdlibm/

[tim-one]

Logged In: YES
user_id=31435

Python code intends to be portable, and the math module
hasn't yet included any platform-specific functions.

I used to write math libraries for a living, so I know about
netlib <wink>. If that's the intended approach to portability ,
it still needs someone to volunteer to do all the non-trivial
coding, testing, doc, license-hassling, and x-platform config
work. Won't be me because I can't make time for it.

Note that you could easily write an extension module
exposing these functions on _your_ platform (assuming your
platform C supplies them).

[rhettinger]

Logged In: YES
user_id=80475

For the reasons mentioned by Tim, closing this RFE.