On 2018/12/04 15:36, Brian Callahan wrote:
>
>
> On 12/2/18 2:34 PM, Charles A Daniels wrote:
> > Thanks to the help of Klemens, Brian, and Stuart, I have ported pcalc
> > (originally written by Peter Glen, now maintained by Mike Frysinger).
> >
> > Brief description:
> >
> > Programmer's calculator, command line utility.
> >
> > There was always a loophole when it came to a need to covert
> > between hexadecimal / decimal / octal / binary.
> >
> > Especially if it involved an operation like 0x1234 + 0x20 It
> > took a lot of hard work, and mostly a good pocket calculator.
> >
> > Main features:
> >
> > o Full math parser, parentheses, add, sub, mult, div,
> > exponential
> > o Automatic conversion between HEX DEC OCT BIN numbers
> > o Mixing different bases in one expression
> > o Definable variables
> > o Math constants (E PI ...)
> > o Built in math functions (sin/cos/sqrt ...)
> >
> > I believe the port is ready for consideration to be added to the ports
> > tree. However, testing on other architectures is is needed - I only have
> > amd64 and macppc machines personally, and part of the test suite fails
> > on macppc (reported upstream: https://github.com/vapier/pcalc/issues/7).
> > If you have a system with another arch, please let me know if `make
> > test` is able to run successfully!
> >
>
> This version is much nicer.
>
> ~Brian
>
Agreed, nearly there. DESCR could do with a little polish though,
I know it's copied from upstream but we can do better :-) loophole
isn't quite the right word (something like "gap in the market"
would fit better, but is a bit clumsy), "covert" should be "convert",
and actually those two short paragraphs are repeating things from
the features list. Maybe something like this?
-------
pcal is a programmer's calculator. It simplifies some operations
(especially working with multiple bases) that are hard work without
this or another programmer's calculator. Main features:
o Full math parser, parentheses, add, sub, mult, div, exponential
o Automatic conversion between HEX DEC OCT BIN numbers
o Mixing different bases in one expression
o Definable variables
o Math constants (E PI ...)
o Built in math functions (sin/cos/sqrt ...)
-------
Re: tests. I don't think it's necessary to patch out the failing tests
(and that may cause problems for updates), I'd prefer to remove the
patch and just add a comment about this.
TEST_TARGET is not needed either, the default works fine, and we can
set TESTSUITEFLAGS= to avoid the failure in "getconf _NPROCESSORS_ONLN".
(Linux getconf requires that this is prefixed with an underscore,
OpenBSD requires that it is not prefixed, and because it's an
extension there's no defined behaviour). e.g.
+# Known failures:
+# tests 001 and 002 fail depending on accuracy of host math lib
+# test 006 fails on 32-bit platforms
+# https://github.com/vapier/pcalc/issues/6 and .../issues/7
TEST_ENV = AUTOCONF_VERSION=${AUTOCONF_VERSION}
-TEST_FLAGS = BINDIR="${TRUEPREFIX}/bin"
-TEST_TARGET = check test
+TEST_FLAGS = BINDIR="${TRUEPREFIX}/bin" TESTSUITEFLAGS=
