Re: [Boost-users] [intrusive] first impression and constant-time removal

Subject: Re: [Boost-users] [intrusive] first impression and constant-time removal
From: Erik Cassel (erik_at_[hidden])
Date: 2009-02-05 12:50:05

Zeljko,

Thank you for your reply. I'm glad to learn that intrusive remove() follows
a known convention in std. I also like your "iterator_of" suggestion.

So that leaves me with 2 remaining suggestions, which I'll restate for the
record:

1) Since constant-time removal is a rather common task, shouldn't the docs
have an example of how to do it properly? You have to dig pretty deeply into
the API to figure out a O(0) way to do it. 

2) Wouldn't a wrapper function for constant-time element removal be a good
addition to the library?

Thanks again,

-Erik

-----Original Message-----
From: boost-users-bounces_at_[hidden]
[mailto:boost-users-bounces_at_[hidden]] On Behalf Of Zeljko Vrba
Sent: Thursday, February 05, 2009 7:44 AM
To: boost-users_at_[hidden]
Subject: Re: [Boost-users] [intrusive] first impression and constant-time
removal

On Thu, Feb 05, 2009 at 06:44:19AM -0800, Erik Cassel wrote:
> 
> 1) I found the remove() function's name ambiguous. It isn't clear that the
>
It fully agrees with the use of the remove() name in the standard
<algorithm>
header.  Calling it something else would break the convention and introduce
unnecessary extra learning effort for people acquainted with the standard
library.

I am actually replying to add item 4: iterator_to is an extremely
unintuitive
and little informative name which I'm continually stumbling over.  The
convention is to name the functions according to what they return, so
"to_iterator" or "get_iterator" or "iterator_of"[1] would make much more
sense.

But I guess it's too late to complain about that "bug" now that the library
has
undergone review and gotten accepted...

[1] Note how "iterator_of" is declarative, indicating return value, whereas
"iterator_to" is imperative and leads the reader to expect that something
will
be done with the iterator that is passed in as argument (compare with
functions
of similar names than one encounters, e.g. "fixed_string_to_int", which
takes
string as a source argument; functions returning X are conventionally named
to_X -- e.g., browse through DateTime library [just a random example]).

I see the logic behind a statement like iterator_to(X), but each time I have
to
use that function, I have to stop, think twice and assure myself that the
naming logic really is the opposite of the convention and that the function
really does what I need.

_______________________________________________
Boost-users mailing list
Boost-users_at_[hidden]
http://lists.boost.org/mailman/listinfo.cgi/boost-users