From: David Abrahams (abrahams_at_[hidden])
Date: 2001-04-22 18:42:52
----- Original Message -----
From: "Daryle Walker" <darylew_at_[hidden]>
> I know there's been a lot of stuff going on since I last mentioned
> "dlw_oprs.zip" (at version 15 in the vault), but does everyone think it's
> good to go? Are there any remaining big problems?
> For those that have come in late, this archive contains an update to
> <boost/operators.hpp> that has:
> + New operator templates for shifting, unary+, etc.
> + New templates the compose other templates together (- and +)
> + Re-implemented old grouped templates in terms of the new ones
> + Added iterator helper templates for input and output iterators
> + Completely redone documentation (requires HTML 4)
I think this last point may be a problem. Last time I checked, many
Unix/Linux users have only older versions of Netscape which don't support
things like “ and ”.
↔ is unknown to IE5 for Windows. I think we may have to stick to HTML
versions < 4.0 for the time being.
The documentation looks quite good, with the following noticeable problems
> Summary of Template Semantics
> Each operator class template completes the concept(s) it describes by
> in operators missing from a target class. A template can implement its new
> operators by using existing base class operations (which would be
> required). Specifically:
What does "which would be required" mean above?
This entire section is good, but needs some English usage cleanup. I would
be happy to do that myself.
The section titled "Use of Concepts" contains the word "expect" where
"except" was intended, I expect ;-)
> The user should be aware of the following points for using this library:
This single body sentence for the "Usage" section adds nothing and should be
The link to the GOTW site needs to be updated. It still works, if you wait
long enough, but I get the feeling that will change eventually.
> Seperate, Explicit Instantiation
^^^^^^^^ should be "separate"
I'm unhappy with the way the tables show up in my browser. The middle column
tends to wrap unneccessarily in every row, causing lots of vertical bloat. I
think this could be fixed by forcing some line breaks in the rightmost
column of the xxx_convertible rows, but probably an HTML expert like you
knows a better way.
I realize we've gone back and forth over this name, but with fresh eyes I
find "negatable" confusible with the idea of unary operator-(). I also buy
Alexy's argument that it is a completely useless template that adds no
functionality. I would like to see it removed.
Overally, I'm very happy with the documentation.
It's pretty close, now, Daryle!
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk