Boost logo

Boost :

From: Reece Dunn (msclrhd_at_[hidden])
Date: 2004-10-23 04:25:21


Bonjour!

Hubert Holin wrote:
> Somewhere in the E.U., le 22/10/2004
> I just wanted to point out that the reason I was interested in it
> is that I had been told (when it was in the sandbox, quite some time
> before the review) it had the potential to help me simplify tremendously
> my quaternion and, especially, octonion libraries, by providing a better
> handling of input operators. So I would say (better late than never,
> though perhaps it is now *too* late) that there actually *is* an
> interest in input facilities.

I haven't yet removed the input facilities from the library. To me,
these are the reasons to keep the input in the library:

[1] The library provides stream operations on STL containers, so why
only provide output of a list/vector/...? This would mean that you'd
need to implement the input operations using some other library, adding
a further dependency on that library.

[2] Symmetry and completeness. I know my library during the review was
called "Output Formatters" (I would like it to be in the
mini/second-review as "I/O Collection Framework" or something similar),
but not providing input would be, IMHO, an oversight.

[3] There have been people in the for input facilities camp as well as
those against. For example, Roland Richter is using my library in
conjunction with lexical_cast to convert between strings and containers.

Therefore, I would prefer to keep the input facilities (at least until
the end of the mini/second-review).

I will commit the updated library once I have finished revising it,
based on the review feedback.

Regards,
Reece


Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk