Board index » cppbuilder » Re: C++Builder Open Letter status

Re: C++Builder Open Letter status


2003-10-07 08:34:21 PM
cppbuilder3
Quote
>"Edward Diener" < XXXX@XXXXX.COM >wrote:
>Hendrik Schober wrote:
>I still don't believe that one needs to know the inner workings of
>template metaprogramming to use it effectively.
"Hendrik Schober" < XXXX@XXXXX.COM >wrote in
Quote
And I still disagree.
And I agree with Edward on this one, if the library is written well <g>
With a well specified (and documented!) public interface the novice user
will not even be aware they are relying on it! [Modulo earlier comments on
error messages]
Quote
>I have easily used boost::function and boost::bind by the book and
>have never had any problems with either.
I havn't used either one, so I cannot comment on them.
Coming to a standard near you soon! You should really check them out. I
find bind in particular to be extremely useful, far suprior to the
std::bind1st/bind2nd it replaces. For my money, bind is the second most
useful library in boost, right after the smart pointers. [Although I find
format to be rather nice as well]
They also illustrate well Edward's point about not requiring deep knowledge
of metaprogramming to make use of it.
AlisdairM
 
 

Re:Re: C++Builder Open Letter status

Hendrik Schober wrote:
Quote
"Edward Diener" < XXXX@XXXXX.COM >wrote:
>Hendrik Schober wrote:
>[...]
>>I haven't used either one, but from my experience
>>with template meta stuff (including my own), I
>>firmly believe that you need to know about the
>>underlying stuff just in order to not to become
>>insane when you misuse something and the compiler
>>goes belly-up with a 47kB error message.
>
>4) Compilers are becoming much more standards conformant so that if
>this problem does occur it may well be the implementation.

What's "this problem"? If I instanciate a
template with a type it isn't supposed to
work with, what help can I get from the
compiler?
If you know the type isn't supported, then don't use it. If you don't know
if the type is supported, then the compiler error message probably means
that it isn't, or that you need to add something to your type so that it
works. Good docs should tell you what that latter addition needs to be. What
I am mainly saying is that if the docs are good, it should tell you how to
use the library and what types the templates accept. If the docs aren't
good, my own personal bias is simply not to use the functionality unless I
understand it pretty completely anyway.
 

Re:Re: C++Builder Open Letter status

"Edward Diener" < XXXX@XXXXX.COM >wrote:
Quote
[...]
>What's "this problem"? If I instanciate a
>template with a type it isn't supposed to
>work with, what help can I get from the
>compiler?

If you know the type isn't supported, then don't use it. If you don't know
if the type is supported, then the compiler error message probably means
that it isn't, or that you need to add something to your type so that it
works. [...]
It might be that I'm just another bad prgogrammer,
but I found over the time that I again and again
have to accept that something either doesn't work
which I thought should or it doesn't work the way
I thought it should. With good error messages from
the compiler, it is usually easier for me to find
out where I failed. With a bad error message it
usually takes quite a while longer for me to find
that out.
Schobi
--
XXXX@XXXXX.COM is never read
I'm Schobi at suespammers org
"And why should I know better by now/When I'm old enough not to?"
Beth Orton
 

{smallsort}

Re:Re: C++Builder Open Letter status

"AlisdairM" < XXXX@XXXXX.COM >wrote:
Quote
[...]
>>I still don't believe that one needs to know the inner workings of
>>template metaprogramming to use it effectively.
[...]
>And I still disagree.

And I agree with Edward on this one, if the library is written well <g>
With a well specified (and documented!) public interface the novice user
will not even be aware they are relying on it! [Modulo earlier comments on
error messages]
See my reply to Edward.
Quote
>>I have easily used boost::function and boost::bind by the book and
>>have never had any problems with either.

>I havn't used either one, so I cannot comment on them.

Coming to a standard near you soon! You should really check them out. I
find bind in particular to be extremely useful, far suprior to the
std::bind1st/bind2nd it replaces. [...]
That's probably the problem. I don't think I
ever used those.
Quote
AlisdairM
Schobi
--
XXXX@XXXXX.COM is never read
I'm Schobi at suespammers org
"And why should I know better by now/When I'm old enough not to?"
Beth Orton
 

Re:Re: C++Builder Open Letter status

Quote
"AlisdairM" < XXXX@XXXXX.COM >wrote:
>Coming to a standard near you soon! You should really check them
>out. I find bind in particular to be extremely useful, far suprior
>to the std::bind1st/bind2nd it replaces. [...]
"Hendrik Schober" < XXXX@XXXXX.COM >wrote in
Quote
That's probably the problem. I don't think I
ever used those.
That's why they're being upgraded with something more useful ;?)
Also check out boost::mem_fn, incredibly useful for iterating over
containers of objects (or reference/[smart]pointer to object)
AlisdairM
 

Re:Re: C++Builder Open Letter status

On Thu, 02 Oct 2003 16:57:45 -0700, John Kaster (Borland) wrote:
Quote
I've got good news and bad news.

First, the bad news: the open letter will not be ready to be posted this
week, as previously promised.

Second, the good news: the open letter will be posted on or by October
17th, 2003.

<SNIP>
Well, it's now October 18 2003, and no letter has appeared yet on
bdn.borland.com. John, can you tell us if the letter is posted
yet, and if so, where it is posted? If not, can you tell us why it is
delayed (again)?
mr_organic
 

Re:Re: C++Builder Open Letter status

"mr_organic" < XXXX@XXXXX.COM >wrote in message
Quote
Well, it's now October 18 2003, and no letter has appeared
yet on bdn.borland.com. John, can you tell us if the letter
is posted yet
It is not.
Quote
If not, can you tell us why it is delayed (again)?
John Kaster and JT have already explained that, if you had read the other
messages in this group. Borland's legal department has looked at the letter
and decided to delay its posting for a few more days pending approval of
some of the details contained within the letter.
Gambit