How many of you have read through the C++ spec, like, all the
way through?

Have you guys checked out the latest version of the C++ spec yet?

I was just browsing around and I stumbled upon this little
guide in the WHATWG HTML spec on how to properly read through
these kinds of technical specifications!

|1.9.1 How to read this specification
| |This specification should be read like all other
|specifications. First, it should be read cover-to-cover,
|multiple times. Then, it should be read backwards at least
|once. Then it should be read by picking random sections from
|the contents list and following all the crossreferences.
| HTML Living Standard - Last Updated 9 April 2024

On 04/22/2024 07:59 AM, Stefan Ram wrote:
> How many of you have read through the C++ spec, like, all the
> way through?
>
> Have you guys checked out the latest version of the C++ spec yet?
>
> I was just browsing around and I stumbled upon this little
> guide in the WHATWG HTML spec on how to properly read through
> these kinds of technical specifications!
>
> |1.9.1 How to read this specification
> |
> |This specification should be read like all other
> |specifications. First, it should be read cover-to-cover,
> |multiple times. Then, it should be read backwards at least
> |once. Then it should be read by picking random sections from
> |the contents list and following all the crossreferences.
> |
> HTML Living Standard - Last Updated 9 April 2024
>

That's great.

I got a lot out of reading "the book", Stroustrup's, then
it reminds me of Harbison and Steele C/C++, and Schildt, the
"Effective C++", some good code with smart pointers,
the I/O streams book or Langer and Kreft, bunch of
Win32 then into COM/DCOM if because, then sort of
about m4 and C/C++ then getting into when .h went to .hpp
then there's much to learn about copy and move and the
in-place and a bunch of sorts great things.

(I'm still sort of learning C99 and C++ 98.)

That's great, though, at least "it should be read".

Ross Finlayson <ross.a.finlayson@gmail.com> wrote:
> On 04/22/2024 07:59 AM, Stefan Ram wrote:
>> How many of you have read through the C++ spec, like, all the
>> way through?
>>
>> Have you guys checked out the latest version of the C++ spec yet?
>>
>> I was just browsing around and I stumbled upon this little
>> guide in the WHATWG HTML spec on how to properly read through
>> these kinds of technical specifications!
>>
>> |1.9.1 How to read this specification
>> |
>> |This specification should be read like all other
>> |specifications. First, it should be read cover-to-cover,
>> |multiple times. Then, it should be read backwards at least
>> |once. Then it should be read by picking random sections from
>> |the contents list and following all the crossreferences.
>> |
>> HTML Living Standard - Last Updated 9 April 2024
>>
>
> That's great.
>
> I got a lot out of reading "the book", Stroustrup's, then
> it reminds me of Harbison and Steele C/C++, and Schildt, the
> "Effective C++", some good code with smart pointers,
> the I/O streams book or Langer and Kreft, bunch of
> Win32 then into COM/DCOM if because, then sort of
> about m4 and C/C++ then getting into when .h went to .hpp
> then there's much to learn about copy and move and the
> in-place and a bunch of sorts great things.
>
> (I'm still sort of learning C99 and C++ 98.)
>
> That's great, though, at least "it should be read".

If you are into component, you can easily skip (D)COM and look into Managed
C++ (C++/CLI) instead. This makes your component available to almost as
many applications as with COM and isn’t burdened by outdated reference
counting. Unless you have to wrestle with DirectX, which AFAIK still
requires COM (another reason to use OpenGL or a more modern framework
instead).

Regards,
Stuart

On 23/04/2024 04:18, Ross Finlayson wrote:
> On 04/22/2024 07:59 AM, Stefan Ram wrote:
>>    How many of you have read through the C++ spec, like, all the
>>    way through?
>>
>>    Have you guys checked out the latest version of the C++ spec yet?
>>
>>    I was just browsing around and I stumbled upon this little
>>    guide in the WHATWG HTML spec on how to properly read through
>>    these kinds of technical specifications!
>>
>> |1.9.1 How to read this specification
>> |
>> |This specification should be read like all other
>> |specifications. First, it should be read cover-to-cover,
>> |multiple times. Then, it should be read backwards at least
>> |once. Then it should be read by picking random sections from
>> |the contents list and following all the crossreferences.
>> |
>> HTML Living Standard - Last Updated 9 April 2024
>>

A much better choice than reading the C++ standards, for most purposes,
is the web site <https://en.cppreference.com/w/>. It is strongly
associated with the ISO working group for C++, and about as close to an
official reference site for C and C++ (core languages and standard
libraries) as you can get. But it is organised as a reference site, not
as a standards document. The site is not a C++ tutorial or something to
read from start to finish, but it is the best place to look things up.

>
> That's great.
>
> I got a lot out of reading "the book", Stroustrup's, then
> it reminds me of Harbison and Steele C/C++, and Schildt, the
> "Effective C++", some good code with smart pointers,
> the I/O streams book or Langer and Kreft, bunch of
> Win32 then into COM/DCOM if because, then sort of
> about m4 and C/C++ then getting into when .h went to .hpp
> then there's much to learn about copy and move and the
> in-place and a bunch of sorts great things.
>
> (I'm still sort of learning C99 and C++ 98.)

Don't bother. That's like saying you want to learn to drive a car, but
are still getting the hang of a horse and cart.

The current C standard is C17, which is just a minor tweak on C11. If
you want to learn C, target C11 - it turned C into a language with
support for multiple threads. Most of the language is the same as C99,
however. (C23 is not yet officially published, AFAIK, even though it
can be considered functionally complete. Compiler support is not
finished yet.)

For C++, things changed so dramatically with C++11 that it is often
considered a new language, and there have been significant improvments
in the language since then. Well-written modern C++ looks as different
from C++98 as it does from C. If you are looking for some kind of
tutorial book, make sure it is based on at least C++17 (and by "based
on", I mean that's the language taught in the book - not just a "What's
new in C++17" appendix add-on).

On 04/23/2024 12:54 AM, David Brown wrote:
> On 23/04/2024 04:18, Ross Finlayson wrote:
>> On 04/22/2024 07:59 AM, Stefan Ram wrote:
>>> How many of you have read through the C++ spec, like, all the
>>> way through?
>>>
>>> Have you guys checked out the latest version of the C++ spec yet?
>>>
>>> I was just browsing around and I stumbled upon this little
>>> guide in the WHATWG HTML spec on how to properly read through
>>> these kinds of technical specifications!
>>>
>>> |1.9.1 How to read this specification
>>> |
>>> |This specification should be read like all other
>>> |specifications. First, it should be read cover-to-cover,
>>> |multiple times. Then, it should be read backwards at least
>>> |once. Then it should be read by picking random sections from
>>> |the contents list and following all the crossreferences.
>>> |
>>> HTML Living Standard - Last Updated 9 April 2024
>>>
>
> A much better choice than reading the C++ standards, for most purposes,
> is the web site <https://en.cppreference.com/w/>. It is strongly
> associated with the ISO working group for C++, and about as close to an
> official reference site for C and C++ (core languages and standard
> libraries) as you can get. But it is organised as a reference site, not
> as a standards document. The site is not a C++ tutorial or something to
> read from start to finish, but it is the best place to look things up.
>
>>
>> That's great.
>>
>> I got a lot out of reading "the book", Stroustrup's, then
>> it reminds me of Harbison and Steele C/C++, and Schildt, the
>> "Effective C++", some good code with smart pointers,
>> the I/O streams book or Langer and Kreft, bunch of
>> Win32 then into COM/DCOM if because, then sort of
>> about m4 and C/C++ then getting into when .h went to .hpp
>> then there's much to learn about copy and move and the
>> in-place and a bunch of sorts great things.
>>
>> (I'm still sort of learning C99 and C++ 98.)
>
> Don't bother. That's like saying you want to learn to drive a car, but
> are still getting the hang of a horse and cart.
>
> The current C standard is C17, which is just a minor tweak on C11. If
> you want to learn C, target C11 - it turned C into a language with
> support for multiple threads. Most of the language is the same as C99,
> however. (C23 is not yet officially published, AFAIK, even though it
> can be considered functionally complete. Compiler support is not
> finished yet.)
>
> For C++, things changed so dramatically with C++11 that it is often
> considered a new language, and there have been significant improvments
> in the language since then. Well-written modern C++ looks as different
> from C++98 as it does from C. If you are looking for some kind of
> tutorial book, make sure it is based on at least C++17 (and by "based
> on", I mean that's the language taught in the book - not just a "What's
> new in C++17" appendix add-on).
>
>

When cppreference arrived it was a great thing,
cross-referenced hyper-linked language, library,
and language level coverage.

I derived value and some entertainment from
Stroustrup's and Sutter's C++ guidelines,
moreso than the Google C++ guidelines.

These days it does look like C++ 11 would be about
the way of things, for somebody last familiar with
"MSVC6 SP3/SP6", though it was great for the Windows
side when they released a C++ compiler, to complement
things like Djgpp, Mingw64, Navia's lcc, and these
kinds of things, vis-a-vis Borland, about a world
where gcc/g++/gpp is a great thing.

On 23/04/2024 17:25, Ross Finlayson wrote:
> On 04/23/2024 12:54 AM, David Brown wrote:
>> On 23/04/2024 04:18, Ross Finlayson wrote:
>>> On 04/22/2024 07:59 AM, Stefan Ram wrote:
>>>>    How many of you have read through the C++ spec, like, all the
>>>>    way through?
>>>>
>>>>    Have you guys checked out the latest version of the C++ spec yet?
>>>>
>>>>    I was just browsing around and I stumbled upon this little
>>>>    guide in the WHATWG HTML spec on how to properly read through
>>>>    these kinds of technical specifications!
>>>>
>>>> |1.9.1 How to read this specification
>>>> |
>>>> |This specification should be read like all other
>>>> |specifications. First, it should be read cover-to-cover,
>>>> |multiple times. Then, it should be read backwards at least
>>>> |once. Then it should be read by picking random sections from
>>>> |the contents list and following all the crossreferences.
>>>> |
>>>> HTML Living Standard - Last Updated 9 April 2024
>>>>
>>
>> A much better choice than reading the C++ standards, for most purposes,
>> is the web site <https://en.cppreference.com/w/>.  It is strongly
>> associated with the ISO working group for C++, and about as close to an
>> official reference site for C and C++ (core languages and standard
>> libraries) as you can get.  But it is organised as a reference site, not
>> as a standards document.  The site is not a C++ tutorial or something to
>> read from start to finish, but it is the best place to look things up.
>>
>>>
>>> That's great.
>>>
>>> I got a lot out of reading "the book", Stroustrup's, then
>>> it reminds me of Harbison and Steele C/C++, and Schildt, the
>>> "Effective C++", some good code with smart pointers,
>>> the I/O streams book or Langer and Kreft, bunch of
>>> Win32 then into COM/DCOM if because, then sort of
>>> about m4 and C/C++ then getting into when .h went to .hpp
>>> then there's much to learn about copy and move and the
>>> in-place and a bunch of sorts great things.
>>>
>>> (I'm still sort of learning C99 and C++ 98.)
>>
>> Don't bother.  That's like saying you want to learn to drive a car, but
>> are still getting the hang of a horse and cart.
>>
>> The current C standard is C17, which is just a minor tweak on C11.  If
>> you want to learn C, target C11 - it turned C into a language with
>> support for multiple threads.  Most of the language is the same as C99,
>> however.  (C23 is not yet officially published, AFAIK, even though it
>> can be considered functionally complete.  Compiler support is not
>> finished yet.)
>>
>> For C++, things changed so dramatically with C++11 that it is often
>> considered a new language, and there have been significant improvments
>> in the language since then.  Well-written modern C++ looks as different
>> from C++98 as it does from C.  If you are looking for some kind of
>> tutorial book, make sure it is based on at least C++17 (and by "based
>> on", I mean that's the language taught in the book - not just a "What's
>> new in C++17" appendix add-on).
>>
>>
>
> When cppreference arrived it was a great thing,
> cross-referenced hyper-linked language, library,
> and language level coverage.
>

I don't know why you are writing in the past tense - yes, the
cppreference website was a great resource when it was first published.
It is /still/ a great resource, and I don't know of any that is better
as a reference site. You make it sound like this is no longer the case.

> I derived value and some entertainment from
> Stroustrup's and Sutter's C++ guidelines,
> moreso than the Google C++ guidelines.

These are all good resources. Different people have different
requirements and preferences, and so will like different sets of guidelines.

>
> These days it does look like C++ 11 would be about
> the way of things, for somebody last familiar with
> "MSVC6 SP3/SP6", though it was great for the Windows
> side when they released a C++ compiler, to complement
> things like Djgpp, Mingw64, Navia's lcc, and these
> kinds of things, vis-a-vis Borland, about a world
> where gcc/g++/gpp is a great thing.
>
>

I'm sorry, I have very little idea what you are trying to say here. It
looks like you are saying that you were happy when MS first made a C++
compiler, and therefore you want to learn an old version of C++. But I
presume I misunderstood, since that would be an irrational argument.

C++11 was the /beginning/ of the new C++ language. It has got better
since then, in many ways. Many things that were hard to write in older
C++ have got progressively easier with new versions. For example, you
can pretty much forget SFINAE and enable_if<>, and their associated
incomprehensible error messages - use concepts in C++20.

There are still bits to come. Contracts will be the big step in C++26.
Metaclasses and reflection will, I hope, make it into C++ in the future,
and hopefully also zero-overhead exceptions.

The disadvantage of backwards compatibility is that new features come in
addition to existing ones, making the language bigger, but for your own
code you can concentrate on using new and better techniques. (That does
not mean you have to like or use /everything/ in newer standards!)

If you are learning a new language, and don't need to do so specifically
to support old code, then it makes sense to learn the /current/ language
and look forward to what's coming - not to learn outdated versions and
look backwards.

On 04/24/2024 12:09 AM, David Brown wrote:
> On 23/04/2024 17:25, Ross Finlayson wrote:
>> On 04/23/2024 12:54 AM, David Brown wrote:
>>> On 23/04/2024 04:18, Ross Finlayson wrote:
>>>> On 04/22/2024 07:59 AM, Stefan Ram wrote:
>>>>> How many of you have read through the C++ spec, like, all the
>>>>> way through?
>>>>>
>>>>> Have you guys checked out the latest version of the C++ spec yet?
>>>>>
>>>>> I was just browsing around and I stumbled upon this little
>>>>> guide in the WHATWG HTML spec on how to properly read through
>>>>> these kinds of technical specifications!
>>>>>
>>>>> |1.9.1 How to read this specification
>>>>> |
>>>>> |This specification should be read like all other
>>>>> |specifications. First, it should be read cover-to-cover,
>>>>> |multiple times. Then, it should be read backwards at least
>>>>> |once. Then it should be read by picking random sections from
>>>>> |the contents list and following all the crossreferences.
>>>>> |
>>>>> HTML Living Standard - Last Updated 9 April 2024
>>>>>
>>>
>>> A much better choice than reading the C++ standards, for most purposes,
>>> is the web site <https://en.cppreference.com/w/>. It is strongly
>>> associated with the ISO working group for C++, and about as close to an
>>> official reference site for C and C++ (core languages and standard
>>> libraries) as you can get. But it is organised as a reference site, not
>>> as a standards document. The site is not a C++ tutorial or something to
>>> read from start to finish, but it is the best place to look things up.
>>>
>>>>
>>>> That's great.
>>>>
>>>> I got a lot out of reading "the book", Stroustrup's, then
>>>> it reminds me of Harbison and Steele C/C++, and Schildt, the
>>>> "Effective C++", some good code with smart pointers,
>>>> the I/O streams book or Langer and Kreft, bunch of
>>>> Win32 then into COM/DCOM if because, then sort of
>>>> about m4 and C/C++ then getting into when .h went to .hpp
>>>> then there's much to learn about copy and move and the
>>>> in-place and a bunch of sorts great things.
>>>>
>>>> (I'm still sort of learning C99 and C++ 98.)
>>>
>>> Don't bother. That's like saying you want to learn to drive a car, but
>>> are still getting the hang of a horse and cart.
>>>
>>> The current C standard is C17, which is just a minor tweak on C11. If
>>> you want to learn C, target C11 - it turned C into a language with
>>> support for multiple threads. Most of the language is the same as C99,
>>> however. (C23 is not yet officially published, AFAIK, even though it
>>> can be considered functionally complete. Compiler support is not
>>> finished yet.)
>>>
>>> For C++, things changed so dramatically with C++11 that it is often
>>> considered a new language, and there have been significant improvments
>>> in the language since then. Well-written modern C++ looks as different
>>> from C++98 as it does from C. If you are looking for some kind of
>>> tutorial book, make sure it is based on at least C++17 (and by "based
>>> on", I mean that's the language taught in the book - not just a "What's
>>> new in C++17" appendix add-on).
>>>
>>>
>>
>> When cppreference arrived it was a great thing,
>> cross-referenced hyper-linked language, library,
>> and language level coverage.
>>
>
> I don't know why you are writing in the past tense - yes, the
> cppreference website was a great resource when it was first published.
> It is /still/ a great resource, and I don't know of any that is better
> as a reference site. You make it sound like this is no longer the case.
>
>> I derived value and some entertainment from
>> Stroustrup's and Sutter's C++ guidelines,
>> moreso than the Google C++ guidelines.
>
> These are all good resources. Different people have different
> requirements and preferences, and so will like different sets of
> guidelines.
>
>>
>> These days it does look like C++ 11 would be about
>> the way of things, for somebody last familiar with
>> "MSVC6 SP3/SP6", though it was great for the Windows
>> side when they released a C++ compiler, to complement
>> things like Djgpp, Mingw64, Navia's lcc, and these
>> kinds of things, vis-a-vis Borland, about a world
>> where gcc/g++/gpp is a great thing.
>>
>>
>
> I'm sorry, I have very little idea what you are trying to say here. It
> looks like you are saying that you were happy when MS first made a C++
> compiler, and therefore you want to learn an old version of C++. But I
> presume I misunderstood, since that would be an irrational argument.
>
> C++11 was the /beginning/ of the new C++ language. It has got better
> since then, in many ways. Many things that were hard to write in older
> C++ have got progressively easier with new versions. For example, you
> can pretty much forget SFINAE and enable_if<>, and their associated
> incomprehensible error messages - use concepts in C++20.
>
> There are still bits to come. Contracts will be the big step in C++26.
> Metaclasses and reflection will, I hope, make it into C++ in the future,
> and hopefully also zero-overhead exceptions.
>
> The disadvantage of backwards compatibility is that new features come in
> addition to existing ones, making the language bigger, but for your own
> code you can concentrate on using new and better techniques. (That does
> not mean you have to like or use /everything/ in newer standards!)
>
> If you are learning a new language, and don't need to do so specifically
> to support old code, then it makes sense to learn the /current/ language
> and look forward to what's coming - not to learn outdated versions and
> look backwards.
>
>
>

Ah. The cppreference site is pretty great, when it came along it was
the greatest bit of organized documentation and nicely is. I just
kind of remember when it came along or started appearing in search
results. It seems gratifying that cppreference makes available an
offline edition. https://en.cppreference.com/w/Cppreference:Archives

About Microsoft and a C++ compiler, what I refer to is when they
released a free-as-in-beer compiler, Express, even just the compiler
itself or cl.exe, because, it was a significant bar to entry of
development on Windows, in C/C++. Then it was that, for a time, for a
while, the "LTS" or "Long Term Support" version of MSVC, was MSVC6, then
as with SP3, then as with SP6, the end. I.e. that was, past tense, the
toolchain, for reproducible builds. (I haven't worked in Windows since
Windows 7, which is 64-bit, vis-a-vis "WoW64" and this kind of thing,
about Windows or Win32 development and the land of ATL/COM, which these
days is "it's always been 64-bit".)

So, as an old duffer, I'm excited about learning C++ 11, figuring
that I can un-learn some things about RVO and some other things
that I wish I never knew, while at the same time, expecting all
of C99/C++98 to compile the same. Like, where placement-new
should be sort of great with MoveConstructible, I wonder.

RTTI is about my favorite thing, and RAII is sort of the very
definition of correctness, initialization, usable objects,
and exceptions are the only way to unwind the stack,
in any case other than many happy returns.

Mostly looking at some Internet Protocols, figuring to implement
some servers in C/C++, it's agreeable that something like
a ready offline edition of the language and libraries is a great thing.
.... To go along with the IETF RFC's, and a POSIX/Windows compatibility
layer mapping (for systems code, while Windows of course has its own
runtime with respect to all the COM components and WMI everything).
Then, I suppose one must eventually include Boost for some reason, yet,
Boost is grand, for piecemeal Boost.

Then there's also much respect for Doxygen.

ram@zedat.fu-berlin.de (Stefan Ram) writes:

> How many of you have read through the C++ spec, like, all the
> way through?
>
> Have you guys checked out the latest version of the C++ spec yet?
>
> I was just browsing around and I stumbled upon this little
> guide in the WHATWG HTML spec on how to properly read through
> these kinds of technical specifications!
>
> |1.9.1 How to read this specification
> |
> |This specification should be read like all other
> |specifications. First, it should be read cover-to-cover,
> |multiple times. Then, it should be read backwards at least
> |once. Then it should be read by picking random sections from
> |the contents list and following all the crossreferences.
> |
> HTML Living Standard - Last Updated 9 April 2024

Any document that includes such advice is giving a tacit
acknowledgement that what is being described is poorly
designed or that the describing document is poorly written,
or both.