cppcms-users

[Cppcms-users] Read-only context

[Nazım C. B. <naz...@ne...>]

Hi Artyom,

First of all, thank you (and all other contributors) for developing such 
good C++ web development library.

Now, in one of application I am developing; I need to pass HTTP context 
in read-only manner. What I am trying to achieve is; writing predicate 
functions taking const reference to context, and checking some 
conditions related with the request. With current version of 
http::context and application classes, this is not possible out-of-box.

That's why I attached a patch to add "const-correct" member functions; 
which works well for my case. If it's possible, could you apply patch to 
trunk? I prefer to not fork a project just for small changes.

Regards,
Nazim.

Re: [Cppcms-users] Read-only context

[Artyom B. <art...@gm...>]

Hello Nazim,

Ohhh, I wish you were talking about it before you started to do all
the hard work.

The const correctness was discussed several times in the list and I
explained why central classes do not provide const interface.

Many of the interfaces that are used relay on lazy
initialization/evaluation - as you noticed for example for

- cache() object - it is created on demand.
- map<string,string> http::request::getenv() does similar stuff - it
converts the efficient internal representation to external only on
demand.

Even if some of the interfaces can be converted to "const" ones I
still prefer to keep an option to use lazy initialization if I will
need
in future to without breaking API or const correctness and without
creating unexpected side effects.

It is very critical as CppCMS is strongly performance oriented and I
need to keep the option for lazy evaluation performance improvements
under the hood.

So thank you for the effort but I can't accept this patch as it is today.

Thanks Again,
  Artyom Beilis


On Tue, Jan 17, 2017 at 6:42 PM, Nazım Can Bedir
<naz...@ne...> wrote:
> Hi Artyom,
>
> First of all, thank you (and all other contributors) for developing such
> good C++ web development library.
>
> Now, in one of application I am developing; I need to pass HTTP context in
> read-only manner. What I am trying to achieve is; writing predicate
> functions taking const reference to context, and checking some conditions
> related with the request. With current version of http::context and
> application classes, this is not possible out-of-box.
>
> That's why I attached a patch to add "const-correct" member functions; which
> works well for my case. If it's possible, could you apply patch to trunk? I
> prefer to not fork a project just for small changes.
>
> Regards,
> Nazim.
>
> ------------------------------------------------------------------------------
> Check out the vibrant tech community on one of the world's most
> engaging tech sites, SlashDot.org! http://sdm.link/slashdot
> _______________________________________________
> Cppcms-users mailing list
> Cpp...@li...
> https://lists.sourceforge.net/lists/listinfo/cppcms-users
>

Re: [Cppcms-users] Read-only context

[Nazım C. B. <naz...@ne...>]

Hello Artyom,

No problem at all.

Thanks,
Nazim.


On 18/01/17 00:00, Artyom Beilis wrote:
> Hello Nazim,
>
> Ohhh, I wish you were talking about it before you started to do all
> the hard work.
>
> The const correctness was discussed several times in the list and I
> explained why central classes do not provide const interface.
>
> Many of the interfaces that are used relay on lazy
> initialization/evaluation - as you noticed for example for
>
> - cache() object - it is created on demand.
> - map<string,string> http::request::getenv() does similar stuff - it
> converts the efficient internal representation to external only on
> demand.
>
> Even if some of the interfaces can be converted to "const" ones I
> still prefer to keep an option to use lazy initialization if I will
> need
> in future to without breaking API or const correctness and without
> creating unexpected side effects.
>
> It is very critical as CppCMS is strongly performance oriented and I
> need to keep the option for lazy evaluation performance improvements
> under the hood.
>
> So thank you for the effort but I can't accept this patch as it is today.
>
> Thanks Again,
>    Artyom Beilis
>
>
> On Tue, Jan 17, 2017 at 6:42 PM, Nazım Can Bedir
> <naz...@ne...> wrote:
>> Hi Artyom,
>>
>> First of all, thank you (and all other contributors) for developing such
>> good C++ web development library.
>>
>> Now, in one of application I am developing; I need to pass HTTP context in
>> read-only manner. What I am trying to achieve is; writing predicate
>> functions taking const reference to context, and checking some conditions
>> related with the request. With current version of http::context and
>> application classes, this is not possible out-of-box.
>>
>> That's why I attached a patch to add "const-correct" member functions; which
>> works well for my case. If it's possible, could you apply patch to trunk? I
>> prefer to not fork a project just for small changes.
>>
>> Regards,
>> Nazim.
>>
>> ------------------------------------------------------------------------------
>> Check out the vibrant tech community on one of the world's most
>> engaging tech sites, SlashDot.org! http://sdm.link/slashdot
>> _______________________________________________
>> Cppcms-users mailing list
>> Cpp...@li...
>> https://lists.sourceforge.net/lists/listinfo/cppcms-users
>>
> ------------------------------------------------------------------------------
> Check out the vibrant tech community on one of the world's most
> engaging tech sites, SlashDot.org! http://sdm.link/slashdot
> _______________________________________________
> Cppcms-users mailing list
> Cpp...@li...
> https://lists.sourceforge.net/lists/listinfo/cppcms-users

Re: [Cppcms-users] Read-only context

[Joerg S. <jo...@be...>]

On Tue, Jan 17, 2017 at 11:00:36PM +0200, Artyom Beilis wrote:
> Many of the interfaces that are used relay on lazy
> initialization/evaluation - as you noticed for example for
> 
> - cache() object - it is created on demand.
> - map<string,string> http::request::getenv() does similar stuff - it
> converts the efficient internal representation to external only on
> demand.

That doesn't necessarily stop const accessors from working, it just
means that the internal fields have to be marked explicitly as mutable.

Joerg

Re: [Cppcms-users] Read-only context

[Nazım C. B. <naz...@ne...>]

Joerg,

I am not expert, but while preparing patch; especially in 'session()' 
and 'cache()' functions; it's not so easy to mark some fields as mutable 
as we could do it in daily C++. Since CppCMS is trying to keep ABI 
stability; it uses pimpl idiom nearly everywhere. And it needs good 
amount of work for refactoring to support mutable in proper way.

(I guess, sort of)

Regards,
Nazim.

On 18/01/17 00:17, Joerg Sonnenberger wrote:
> On Tue, Jan 17, 2017 at 11:00:36PM +0200, Artyom Beilis wrote:
>> Many of the interfaces that are used relay on lazy
>> initialization/evaluation - as you noticed for example for
>>
>> - cache() object - it is created on demand.
>> - map<string,string> http::request::getenv() does similar stuff - it
>> converts the efficient internal representation to external only on
>> demand.
> That doesn't necessarily stop const accessors from working, it just
> means that the internal fields have to be marked explicitly as mutable.
>
> Joerg
>
> ------------------------------------------------------------------------------
> Check out the vibrant tech community on one of the world's most
> engaging tech sites, SlashDot.org! http://sdm.link/slashdot
> _______________________________________________
> Cppcms-users mailing list
> Cpp...@li...
> https://lists.sourceforge.net/lists/listinfo/cppcms-users

Re: [Cppcms-users] Read-only context

[Joerg S. <jo...@be...>]

On Wed, Jan 18, 2017 at 02:25:47AM +0300, Nazım Can Bedir wrote:
> Joerg,
> 
> I am not expert, but while preparing patch; especially in 'session()' 
> and 'cache()' functions; it's not so easy to mark some fields as mutable 
> as we could do it in daily C++. Since CppCMS is trying to keep ABI 
> stability; it uses pimpl idiom nearly everywhere. And it needs good 
> amount of work for refactoring to support mutable in proper way.

On the contrary, that actually tends to make it easier.

Joerg