I would like to get feedback on following operator templates which I am polishing for boost. The purpose of these templates is to define global operators in...
177
Sean A Corfield
sean@...
Jul 1, 1999 9:18 am
Beman Dawes <beman@...> wrote: [One concern is that the message written to cout by the job_timer class] I can't believe you've tied a class to cout! The...
178
Valentin Bonnard
Bonnard.V@...
Jul 1, 1999 9:36 am
... I have written such classes but with just one type T. (See http://www.eleves.ens.fr:8080/home/bonnard/NewLibs/Utils/comparison_ops.H) I have other similar...
179
Valentin Bonnard
Bonnard.V@...
Jul 1, 1999 9:46 am
... Why ? ... Like: http://www.eleves.ens.fr:8080/home/bonnard/agil/item.html#ProgressIndicator Note that this base class is by definition a DialogItem so it's...
180
Dave Abrahams
abrahams@...
Jul 1, 1999 12:47 pm
... You don't like using multiple-inheritance for TotallyOrdered? I'm not sure we should have competing submissions in the same domain. It's enough work to...
181
Dave Abrahams
abrahams@...
Jul 1, 1999 12:48 pm
... Because the default assignment semantics are seldom what you want. ... eGroups.com home: http://www.egroups.com/group/boost http://www.egroups.com -...
182
Ed Brey
brey@...
Jul 1, 1999 1:08 pm
... I'm assuming just a cut-and-paste typo, but the about two functions should not be operator-. It would be convenient to add a class that pulls in all the...
183
Valentin Bonnard
Bonnard.V@...
Jul 1, 1999 1:42 pm
... 1) here the code redefine the default assignment semantics 2) there is no default assignment semantics at all -- Valentin Bonnard ... eGroups.com home:...
184
Valentin Bonnard
Bonnard.V@...
Jul 1, 1999 1:45 pm
... 1) unlike some people, I never try to avoid MI (in other words, I use MI when I feel a need for it) 2) I just cannot see how MI would be used here 3) this...
185
Beman Dawes
beman@...
Jul 1, 1999 1:59 pm
... http://www.ozemail.com.au/~getknown/simetric.html#common ... contexts ... nothing ... Thanks to Andy, Ed, and Valentin for their replies. I had never even...
186
Beman Dawes
beman@...
Jul 1, 1999 2:08 pm
... See ... You are right, of course, that logically they are unrelated. The rationale for combining them was: * Avoiding header proliferation. * In...
... class] ... should be a ... timing ... elapsed() ... Seems like there are a couple of issues coming forward for job_timer and progress_display: 1) They are...
189
Beman Dawes
beman@...
Jul 1, 1999 4:24 pm
... doing the ... Ah, yes, that would be better. The code was written last year, before boost had the smart pointers. I will change it. ... looks ... hard to...
190
Beman Dawes
beman@...
Jul 1, 1999 4:24 pm
... the ... up ... What is needed is CLOCK_T_MAX or similar, which may be less than the max value for the underlying type. I have already had a couple of...
191
Nicolai Josuttis
nicolai.josuttis@...
Jul 1, 1999 5:12 pm
... This is to honor the fact that the SGI implementation uses compose1 and compose2 instead of compose_f_gx and compose_f_gx_hx for years now to stay...
192
Nathan Myers
ncm@...
Jul 1, 1999 5:47 pm
... You don't need macros for that. You can provide duplicate definitions, or inlines that call the official names. BTW: Is the sense of the #ifndef, above,...
193
Reid Sweatman
reids@...
Jul 1, 1999 6:26 pm
... But isn't that sort of the architectural version of "optimizing code before finding the best algorithm?" ... eGroups.com home:...
194
Reid Sweatman
reids@...
Jul 1, 1999 6:41 pm
Just a minor cavil from someone who does most of his programming on Wintel platforms: the C and system timer functions aren't very reliable, because they're...
195
Reid Sweatman
reids@...
Jul 1, 1999 6:44 pm
Not to be a pest, but personally, I'd rather see the timer and display functionality separated. Yes, you probably would always use the timer functionality...
196
Ed Brey
brey@...
Jul 1, 1999 7:21 pm
... It is possible to get this flexiblity without sacrificing (i.e. adding) the level of indirection that the reference/pointer causes. Something like this...
197
Beman Dawes
beman@...
Jul 1, 1999 7:45 pm
... Wintel ... because ... in every ... Pentium and ... available, ... since ... instructions are ... handled ... subset ... chunk ... implementation layer ......
198
Reid Sweatman
reids@...
Jul 1, 1999 8:06 pm
Well, the classes I've been using for my own stuff aren't templates, so they're not generic. If that's not a problem, I don't mind showing them. They're a...
199
Nicolai Josuttis
nicolai.josuttis@...
Jul 1, 1999 8:39 pm
... I know (and I don't want to start a war about using macros again). I just used macros here because they were simple one-lines to make a quick hack for...
200
Dietmar Kuehl
dietmar.kuehl@...
Jul 1, 1999 9:16 pm
Hi, ... I think this applies to everything available from the boost pages: The main contribution is the interface which is augmented with an implementation...
201
Dave Abrahams
abrahams@...
Jul 2, 1999 12:47 am
... Can't you do the same thing with an inline forwarding function, wrapper class or some other mechanism that doesn't walk across namespace boundaries the way...
202
Dave Abrahams
abrahams@...
Jul 2, 1999 12:50 am
... Thanks for catching that. ... Good thought. I'll do that. ... eGroups.com home: http://www.egroups.com/group/boost http://www.egroups.com - Simplifying...
203
Andy Glew
glew@...
Jul 2, 1999 1:48 am
As somebody who has worked on the architectural definition of the Intel timestamp counter and RDTSC instructions, mind if I respond/add to Reid's statement...
204
Ed Brey
brey@...
Jul 2, 1999 12:33 pm
... The C++ standard provides a good example of just such a generic approach: allocators. They allow generic routines to work with platform-specific memory...
205
Beman Dawes
beman@...
Jul 2, 1999 6:07 pm
... better? ... more ... I very much agree with Dietmar that this is an important issue for boost.org. I don't know of any single general approach to dealing ...
