Re: Text collation

[Kevin Ryde]

address@hidden (Ludovic Courtès) writes:
>
> So, do you think we should rename them to `strto{d,l}' or documenting
> this dependence is enough?

Number parsing on the whole always seems very application-specific to
me.  I don't think I've ever seen a style that could recommended
whole-heartedly for all user input :-(.

I'd think about something modest like a string->number that stripped
thousands seps and transformed the decimal point (if necessary).  I
think that'd find a use.

> Hmm, I thought `unresolved' was for cases where a test cannot be run for
> some reason.  See, e.g., `guardians.test',

Not sure about those, were they actually unresolved issues for a time?

> `socket.test'

Hang on, that's your work isn't it! :)

> Well, there's admittedly not a huge difference.

Yes, though hopefully the difference is that "unsupported" bits can be
safely ignored, assuming you know those features are not supposed to
be available.  But the author of this stuff must be lurking here ...