john.michael.hall at gmail.com
Wed Feb 26 14:59:31 UTC 2020
On Wednesday, 26 February 2020 at 12:17:43 UTC, Martin Nowak
> Glad to announce the first beta for the 2.091.0 release, ♥ to
> the 55 contributors.
I'm happy to see those Windows install improvements.
In the changelog, it mentions isClose, but I didn't see that
listed in the Phobos enhancement section where it links to
Bugzilla. Maybe there should be a way to link to the PR  in
cases like that? The author addresses two bugzilla issues, but
does not link to them in the PR.
I'm a little confused by the note for isClose where it says that
isClose(a, b, 1e-2, 1e-2) is very similar to approxEqual(a, b).
The default for the second optional parameter for approxEqual is
1e-5, not 1e-2.
Also, there is a typo in the isClose documentation: "This is
mainly usefull for comparing values to zero." Two l's in usefull.
It looks like this typo is also in the approxEqual documentation.
While you're at it, you might make clear that it checks the
relative difference first (unless maxRelDiff=0) and then the
absolute difference (unless maxAbsDiff=0). That made me
realize...for isClose(a, b, 0, 0), it looks like it returns true
if a == b, but false otherwise (ignoring infinity). Is that
intended? I would think that it should instead have an assert say
that the checks are skipped, but that's just me.
More information about the Digitalmars-d-announce