cppcms-users — Disussions of users of CppCMS

[Cppcms-users] cppcms::http::response::finalize() question

[CN <cn...@gr...>]

Hi!

I notice that example code "chat" does not always call  
cppcms::http::response::finalize() before it calls  
async_complete_response().

Will it cause memory leak if cppcms::http::response::finalize() is not  
called before async_complete_response() is called by asynchronous  
applications? If it will, I propose some warnings of its misuse be written  
to the documentation.

Regards,

CN

Re: [Cppcms-users] Run cppcms service as daemon

[CN <cn...@gr...>]

On Wed, 20 Oct 2010 04:51:18 +0800, Artyom  
<art...@ya...> wrote:

> Daemonization is implemented in trunk. Together with:
>
> - Switching to unprivileged user and group
> - Chrooting to specific directry for extra paranoic users.

Would it be even better to also add options "user" and "group" under  
"service.socket" so we can specify something like the following?

service: socket {
	"path": "/tmp/cppcms.socket",
	"user":"www-data",
	"group":"www-data"
}

Regards,

CN

Re: [Cppcms-users] Creating a code-debugging Pastebin with Wt

[Artyom <art...@ya...>]

I don't see any reason why would CppCMS not work for you. Go with it.

About Wt... I have my opinion on this project - I don't think
that entire concept of Wt is correct.

You can see my opinion there:

   http://stackoverflow.com/questions/4157895/web-framework-api/4158839#4158839

That is why I created CppCMS while Wt was around.

If you ask at Wt lists they think they do "the right thing"
while I think that CppCMS does "the right thing"

So...

Its up to you :-)

Regards,
  Artyom


> Hi
> 
> I'm thinking of writing a pastebin with Wt.
> 
> A couple of  questions before I begin;
> 
> • Would you recommend using CppCMS for creating  pastebins?
> • Are there any examples of pastebins created with  CppCMS?
> 
> Please tell me your thoughts on this  project...
> 
> Thanks
> 


      

Re: [Cppcms-users] Creating a code-debugging Pastebin with Wt

[Alec T. <ale...@gm...>]

Apologies fellow CppCMS member, I forgot to change the title and 3rd Line.

Basically, I'm doing research for the C++ web toolkit which will be
suit my needs.

Please reply

On Tue, Nov 16, 2010 at 11:37 PM, Alec Taylor <ale...@gm...> wrote:
> Hi
>
> I'm thinking of writing a pastebin with Wt.
>
> A couple of questions before I begin;
>
> • Would you recommend using CppCMS for creating pastebins?
> • Are there any examples of pastebins created with CppCMS?
>
> Please tell me your thoughts on this project...
>
> Thanks
>
> BTW: I'll be creating one capable of compiling code, akin to codepad
> and IDEone, only with support for multiple files, uploading of files
> and downloading of compiled/outputted files. There will be a complete
> authentication system, however registration will be optional.
>

[Cppcms-users] Creating a code-debugging Pastebin with Wt

[Alec T. <ale...@gm...>]

Hi

I'm thinking of writing a pastebin with Wt.

A couple of questions before I begin;

• Would you recommend using CppCMS for creating pastebins?
• Are there any examples of pastebins created with CppCMS?

Please tell me your thoughts on this project...

Thanks

BTW: I'll be creating one capable of compiling code, akin to codepad
and IDEone, only with support for multiple files, uploading of files
and downloading of compiled/outputted files. There will be a complete
authentication system, however registration will be optional.

Re: [Cppcms-users] std::vector in coding standarts

[Artyom <art...@ya...>]

> 
> Hello Artyom!
> 
> May be it's not so significant, but i think it's necessary  to discuss.
> 
> At  http://cppcms.sourceforge.net/wikipp/en/page/cppcms_1x_coding_standards you  
>wrote that "Always prefer std::vector to std::list - it has better performance  
>because it is cache friendly."
> But imho, doing so can lead to performance  troubles in some cases: for 
>example, if you try to remove an item from the  middle of multimillion 
>std::vector - unlike std::list.
> 
> ---
> Regards,  kpeo


Of course not in case you want to do something like that.

The coding standard is advice, not "always use it" so for example if I do
need to insert items and remove items from list randomly I use std::list,
for example cache where std::list is used for LRU handling.

Artyom


      

[Cppcms-users] std::vector in coding standarts

[<sla...@ya...>]

Hello Artyom!

May be it's not so significant, but i think it's necessary to discuss.

At http://cppcms.sourceforge.net/wikipp/en/page/cppcms_1x_coding_standards you wrote that "Always prefer std::vector to std::list - it has better performance because it is cache friendly."
But imho, doing so can lead to performance troubles in some cases: for example, if you try to remove an item from the middle of multimillion std::vector - unlike std::list.

---
Regards, kpeo

Re: [Cppcms-users] Problem with cppcms::application::add() method

[Artyom <art...@ya...>]

Ok, now I've got it. I just totally misunderstand what you mean, and almost all 
I had written
before is... wrong (except the 3rd problem that this is intentional).

>
>First problem:
>
>void application::add(application &app,std::string regex,int part)
>>{
>>        add(app);
>>        url_dispatcher().mount(regex,app,part);
>>}
>>
> The code above has an error, the correct code is:
> dispatcher().mount(regex,app,part);
>


Fixed. in trunk, changeset 1517


> Second problem:
> 
> When I attach or add an application B to an application A, the method "void 
> B::main(std::string url)" is never called on url dispatch stage.
> 
> This behaviour causes problems when adding cppcms::rpc::json_rpc_server 
> application because the parsing of the request is made in json_rpc_server::main 
>
> method.
> The result is that the application does not work.

You are right! Fixed in trunk changeset 1517


> Third problem:
> 
> in cppcms::rpc::json_rpc_server application: URL mapping with 
> dispatcher().assign() method does not work because main method overwriting.

This is by design, As you actually want to dispatch by RPC function calls 
 rather
then by URL pattern.


Thank you very much for this report and description and Fix you proposed.
This is bad idea to answer on mails in 23:00 :-)

Artyom



      

Re: [Cppcms-users] Problem with cppcms::application::add() method

[Daniel V. <chi...@gm...>]

Hello.

I use the latest cppcms revision.

I answer between the lines below.

Thanks.


On Thu, 2010-11-11 at 12:39 -0800, Artyom wrote:

> >
> >void application::add(application &app,std::string regex,int part)
> >{
> >        add(app);
> >        url_dispatcher().mount(regex,app,part);
> >}
> >
> >The code above has an error, the correct code is:
> >dispatcher().mount(regex,app,part);
> >

You write "url_dispatcher()" in the above method definition.  As you
know  url_dispatcher() makes an anonymous instance of
cppcms::url_dispatcher class. I suspect that your intentions are to call
dispatcher() method instead. 


> 
> add or attach member functions do only one thing - they share the http::context
> class between all connected applications.
> 
> Otherwise you'll just have an exception trying to access anything like request() 
> 
> or response() of any class that wasn't added.
> 
> On the other hand url_dispatching is just a simple way to "split" your main
> function according to requests between its own member function and other sub 
> applications.
> 
> There are different tasks and they both required (unless you wan't to do 
> something on
> your own in main rather then standard URL-dispatching)

They are diferent task, I understand, but  the method add(application
&app,std::string regex,int part) makes these two tasks. 

> 
> > Second problem:
> >
> > When I attach or add an application B to an application A,
> > the method "void B::main(std::string url)" is never 
> > called on url dispatch stage.
> 
> When you mount a sub application you just extract sub expression and
> pass it to mounted application. I.e.
> 
>    add(users_app)
>    add(uploads)
>    urd_dispatched().mount("/users(/.*)",users_app,1);
>    urd_dispatched().mount("/uploads(/.*)",uploads,1);

I can't use that interface because the mount() method is private.
The code

        add(users_app);
        dispatcher().mount("/users(/.*)",u,1);

raise the error

        cppcms/url_dispatcher.h:160: error: ‘void
        cppcms::url_dispatcher::mount(std::string, cppcms::application&,
        int)’ is private
        main.cpp:30: error: within this context

For that reason I use the shortcut method add(application
&app,std::string regex,int part).


> And then in constructor of users you mount
> 
>    url_dispatcher().attach("/login",&users_class::login,this);
> 
> when you hit /users/login
> 
> /users/login is matched against "/users(/.*)" regex and passed to 
> users_app.main()
> with parameter "/login"
> 
> then it matched by users_app's url_disptacher against "/login" regex and passed 
> to login
> member function of users_app instance.
> 
> So you need both steps and understand how they work.
> 
> > This behaviour causes problems when adding cppcms::rpc::json_rpc_server 
> > application because the parsing of the request is made in json_rpc_server::main 
> >
> > method.
> > The result is that the application does not work.
> 
> adding application has nothing to do with URL dispatching, url_disptaching just 
> allows
> you to forward request to other app's main function.

I wrote an small example that reproduces the issue:

        #include <cppcms/application.h>
        #include <cppcms/service.h>
        #include <cppcms/applications_pool.h>
        #include <cppcms/url_dispatcher.h>
        #include <cppcms/http_response.h>
        #include <iostream>
        
        class users : public cppcms::application
        {
        public:
            users(cppcms::service& srv): application(srv) {
                dispatcher().assign("/login$", &users::login, this);
            }
            void main(std::string url) {
                response().out() << "into users::main(\""<< url
        <<"\")\n";
                cppcms::application::main(url);
            }
            void login() {
                response().out() << "into users::login()\n";
            }
        };
        
        class main_app : public cppcms::application
        {
        public:
            main_app(cppcms::service& srv) : application(srv), u(srv) {
                add(u,"/users(/.*)",1);
                //add(u);
                //dispatcher().mount("/users(/.*)",u,1);
            }
            void main(std::string url) {
                response().out() << "into main_app::main(\""<< url
        <<"\")\n";
                cppcms::application::main(url);
            }
        private:
            users u;
        };
        
        int main(int argc, char **argv)
        {
            try {
                cppcms::service srv(argc, argv);
                srv.applications_pool().mount(
                    cppcms::applications_factory<main_app>());
                srv.run();
        
            }
            catch(std::exception const &e) {
                std::cerr << e.what() << std::endl;
                return 1;
            }
            return 0;
        }

Test cases:

A - Test url: .../users/login
     result: "into users::login()"

    That's ok.

B - Test url: .../users/foo
    result: "into main_app::main("/users/foo")"

    That's no ok, because the espected result is "into
users::main("/foo")". The users::main() method is never called.

    Am I wrong?

If you don't change "url_dispatcher()" to "dispatcher()" in
add(application &app,std::string regex,int part) definition then the A
test also fails.



> > in cppcms::rpc::json_rpc_server application: URL mapping with 
> > dispatcher().assign() method does not work because main method overwriting.
> 
> When you develop json_rpc server you suppose to rather bind member function
> by remote procedure functions (i.e. RPC service function = member function)
> rather then using URL-dispatching method suitable for ordinary web applications.
>     
> So it is done by design.
> 
> Artyom
> 
> 
> 
>       
> 
> ------------------------------------------------------------------------------
> Centralized Desktop Delivery: Dell and VMware Reference Architecture
> Simplifying enterprise desktop deployment and management using
> Dell EqualLogic storage and VMware View: A highly scalable, end-to-end
> client virtualization framework. Read more!
> http://p.sf.net/sfu/dell-eql-dev2dev
> _______________________________________________
> Cppcms-users mailing list
> Cpp...@li...
> https://lists.sourceforge.net/lists/listinfo/cppcms-users

Re: [Cppcms-users] Problem with cppcms::application::add() method

[Artyom <art...@ya...>]

>
>void application::add(application &app,std::string regex,int part)
>{
>        add(app);
>        url_dispatcher().mount(regex,app,part);
>}
>
>The code above has an error, the correct code is:
>dispatcher().mount(regex,app,part);
>

add or attach member functions do only one thing - they share the http::context
class between all connected applications.

Otherwise you'll just have an exception trying to access anything like request() 

or response() of any class that wasn't added.

On the other hand url_dispatching is just a simple way to "split" your main
function according to requests between its own member function and other sub 
applications.

There are different tasks and they both required (unless you wan't to do 
something on
your own in main rather then standard URL-dispatching)

> Second problem:
>
> When I attach or add an application B to an application A,
> the method "void B::main(std::string url)" is never 
> called on url dispatch stage.

When you mount a sub application you just extract sub expression and
pass it to mounted application. I.e.

   add(users_app)
   add(uploads)
   urd_dispatched().mount("/users(/.*)",users_app,1);
   urd_dispatched().mount("/uploads(/.*)",uploads,1);

And then in constructor of users you mount

   url_dispatcher().attach("/login",&users_class::login,this);

when you hit /users/login

/users/login is matched against "/users(/.*)" regex and passed to 
users_app.main()
with parameter "/login"

then it matched by users_app's url_disptacher against "/login" regex and passed 
to login
member function of users_app instance.

So you need both steps and understand how they work.

> This behaviour causes problems when adding cppcms::rpc::json_rpc_server 
> application because the parsing of the request is made in json_rpc_server::main 
>
> method.
> The result is that the application does not work.

adding application has nothing to do with URL dispatching, url_disptaching just 
allows
you to forward request to other app's main function.

> in cppcms::rpc::json_rpc_server application: URL mapping with 
> dispatcher().assign() method does not work because main method overwriting.

When you develop json_rpc server you suppose to rather bind member function
by remote procedure functions (i.e. RPC service function = member function)
rather then using URL-dispatching method suitable for ordinary web applications.
    
So it is done by design.

Artyom



      

[Cppcms-users] Problem with cppcms::application::add() method

[Daniel V. <chi...@gm...>]

Hello.

I found problems with cppcms::application::add() and
cppcms::application::attach() methods

First problem:

        void application::add(application &app,std::string regex,int
        part)
        {
                add(app);
                url_dispatcher().mount(regex,app,part);
        }

The code above has an error, the correct code is:

        dispatcher().mount(regex,app,part);




Second problem:

When I attach or add an application B to an application A, the method
"void B::main(std::string url)" is never called on url dispatch stage.

This behaviour causes problems when adding cppcms::rpc::json_rpc_server
application because the parsing of the request is made in
json_rpc_server::main method.
The result is that the application does not work.

Third problem:

in cppcms::rpc::json_rpc_server application: URL mapping with
dispatcher().assign() method does not work because main method
overwriting.
    

Thank you.

Re: [Cppcms-users] Segfaults in libcppcms in asynchronous application

[Artyom <art...@ya...>]

Thanks for the report,

Fixed, take the version from trunk or just pick a changeset 1513 from SVN.

Artyom



----- Original Message ----
> From: Julian Pietron <ju...@wh...>
> To: cpp...@li...
> Sent: Wed, November 10, 2010 3:04:25 PM
> Subject: [Cppcms-users] Segfaults in libcppcms in asynchronous application
> 
> Hallo,
> 
> in an asynchronous application, a segfault is reproducible if a  connection
> of a released context is aborted by browser.
> gdb backtrace is  attached. If you need further information, don't hesitate
> to  ask.
> 
> Thanks for your help,
> Julian
> 


      

Re: [Cppcms-users] Segfaults in libcppcms in asynchronous application

[Artyom <art...@ya...>]

Thanks,

Please open a bug, I'll try to solve it ASAP

Artyom



----- Original Message ----
> From: Julian Pietron <ju...@wh...>
> To: cpp...@li...
> Sent: Wed, November 10, 2010 3:04:25 PM
> Subject: [Cppcms-users] Segfaults in libcppcms in asynchronous application
> 
> Hallo,
> 
> in an asynchronous application, a segfault is reproducible if a  connection
> of a released context is aborted by browser.
> gdb backtrace is  attached. If you need further information, don't hesitate
> to  ask.
> 
> Thanks for your help,
> Julian
> 


      

Re: [Cppcms-users] Segfaults in libcppcms in asynchronous application

[Julian P. <ju...@wh...>]

I forgot to add that I'm using current beta (0.99.4) of libcppcms.

[Cppcms-users] Segfaults in libcppcms in asynchronous application

[Julian P. <ju...@wh...>]

Hallo,

in an asynchronous application, a segfault is reproducible if a connection
of a released context is aborted by browser.
gdb backtrace is attached. If you need further information, don't hesitate
to ask.

Thanks for your help,
Julian

Re: [Cppcms-users] upload progress

[Aris S. <ari...@gm...>]

another solution in the client side is using YUI 3 uploader, an
swfuploader, which have BDS license.

The YUI Uploader leverages Flash to provide file upload functionality
beyond the native browser-based methods. Specifically, the Uploader
allows for:

   1. Multiple file selection in a single "Open File" dialog.
   2. File extension filters to facilitate the user's selection.
   3. Progress tracking for file uploads.
   4. A range of available file metadata: filename, size, date
created, date modified, and author.
   5. A set of events dispatched on various aspects of the file upload
process: file selection, upload progress, upload completion, data
return, and upload errors.
   6. Inclusion of additional data in the file upload POST request.
   7. Faster file upload on broadband connections (due to the modified
SEND buffer size).

for a complete description, read on http://developer.yahoo.com/yui/3/uploader/
for single file upload example, read on
http://developer.yahoo.com/yui/3/examples/uploader/uploader-simple.html#
for multiple file, read on
http://developer.yahoo.com/yui/3/examples/uploader/uploader-multiple.html

another feature of YUI oploader is that "the file upload can be
aborted" with calling "void cancel(fileID)" method, read on
http://www.yuiblog.com/blog/2009/02/26/flickr-uploadr/

for the complete api, read on http://developer.yahoo.com/yui/3/api/Uploader.html

Re: [Cppcms-users] upload progress

[pavel k. <un...@fu...>]

Zheng Ping <kytexzy@...> writes:

> 
> 
> I think swfupload is a good solution.

maybe plupload is better long term solution. did anybody tried to use it with
cppcms? or anything else to handle multiple files upload (with chunking)?

[Cppcms-users] DbiXX updates in trunk

[Artyom <art...@ya...>]

Hello All,

As far as I can see there are some issues and questions about DbiXX,
I updated DbiXX library with different fixes and improvements:


- Fixed time shift when fetching local time during daylight time savings
- Correct overloading for all integer types
- Access to dbi data (dbixx::result)
- Support of single connection string instead of passing multiple parameters
- Support of getting driver name
- Doxygen generated documentation
- Some code cleanup - to make it locale safe

Please take a look on these updates in svn-trunk if you are interested .

Artyom


      

Re: [Cppcms-users] dbixx::result && dbixx::row

[Artyom <art...@ya...>]

> Subject: [Cppcms-users] dbixx::result && dbixx::row
> 
> Hello, is it possible to make dbixx:: result and row serializable? It would  
>let
> us to store db queries results in  cache.
> 
>

This is bad place to do serialization. As dbixx::result are tied to specific DB 
connection
and not generally useful outside this scope.

You need to fetch an object from the database you need using "row" and then 
serialize the
object rather then serializing the row itself.

See for example:
Wikipp's options class:

http://cppcms.svn.sourceforge.net/viewvc/cppcms/wikipp/branches/for_cppcms_v100/options.cpp?revision=1255&view=markup


Function voidoptions::load(), lines 73-90

Artyom


      

Re: [Cppcms-users] getting dbixx::session::escaped_query string

[Artyom <art...@ya...>]

> 
> Can't  dbixx::session::escaped_query be made public?
> That would be useful both for  bebugging and query logging purposes?
> 
> Maybe it would be safer to simply  implement a getEscaped_query() function, but 
>
> how could I use my own function  to modify the query in other ways (e.g. 
> automatically add a prefix to table  names)?

Probably I can add get_escaped_query() member function. And no, I can't make it 
public,
as it private variable used internally.

> Even if I make a class derived from dbixx::session, I still wouldn't  be able 
> to read AND modify the query as long as escaped_query is private,  right?
> 

dbixx::session is not intended to be derived from.


Artyom


      

Re: [Cppcms-users] dbixx::result

[Artyom <art...@ya...>]

Mistake.

Actually these members are for low level access to libdbi for functions missing 
in dbixx.
And generally should not be used.


Artyom


----- Original Message ----
> Hello,
> 
> Why dbixx::result doesn't have "dbi_result get_dbi_result() member 
> function  ?
> 
> Jean-Michel
> 
> 
> 


      

[Cppcms-users] dbixx::result && dbixx::row

[Abramov A. <he...@gm...>]

Hello, is it possible to make dbixx:: result and row serializable? It would let
us to store db queries results in cache.

Re: [Cppcms-users] getting dbixx::session::escaped_query string

[augustin <aug...@ov...>]

On Monday 01 November 2010 03:29:39 pm Artyom wrote:
 
> What is md5 function? Is it cppcms::util::md5 or something else? Because
> of so it creates 16 characters binary string and you probably
> need cppcms::util::md5hex
> for 32 hexadecimal characters.


It was something else that I found elsewhere on the web.
I forgot you already provided the equivalent within the utils.
I now use cppcms::util::md5hex and it works equally well. Thanks.
 
> BTW I suggest to salt passwords to prevent using rainbow tables.

Yes, I will.

> You can add debug printing in line 310 of session.cpp in function
> session::single
> Before line:
> 
>         dbi_result res=dbi_conn_query(conn,escaped_query.c_str());
> 
> Add:
> 
>         std::cerr << "[" << escaped_query <<"]" << std::endl;
> 
> Also if exception is thrown
> you can request query() parameter of dbixx_error.

Can't dbixx::session::escaped_query be made public?
That would be useful both for bebugging and query logging purposes?

Maybe it would be safer to simply implement a getEscaped_query() function, but 
how could I use my own function to modify the query in other ways (e.g. 
automatically add a prefix to table names)?
Even if I make a class derived from dbixx::session, I still wouldn't be able 
to read AND modify the query as long as escaped_query is private, right?




 augustin.




-- 
Friends: http://www.reuniting.info/ http://activistsolutions.org/
My projects:
http://astralcity.org/ http://3enjeux.overshoot.tv/ http://linux.overshoot.tv/ 
http://overshoot.tv/ http://charityware.info/ http://masquilier.org/
http://openteacher.info/ http://minguo.info/ 
http://www.wechange.org/ http://searching911.info/












.

[Cppcms-users] dbixx::result

[Jean-Michel C. <jea...@la...>]

Hello,

Why dbixx::result doesn't have "dbi_result get_dbi_result() member 
function ?

Jean-Michel

Re: [Cppcms-users] dbixx::sql truncates string!

[Artyom <art...@ya...>]

> I get a username and a password from a user and hash the  password using 
md5():
>         std::string pass_md5 =  md5(password);
>         memset (password, '*', 100); //  Erase the password from memory.
>         cout <<  "The md5 is: " << pass_md5 << endl;

What is md5 function? Is it cppcms::util::md5 or something else? Because
of so it creates 16 characters binary string and you probably 
need cppcms::util::md5hex
for 32 hexadecimal characters.

BTW I suggest to salt passwords to prevent using rainbow tables.

>          sql << "SELECT * FROM users WHERE name = ? AND pass = ?", username, 
> pass_md5;
>         if (sql.single(r)) { /* Login  successful...*/    }
> 
> The (critical) problem is that the login  is never successful because in the 
> above query, the pass_md5 is truncated.  The md5 hash is 32 characters long, 
> but only the 20 first characters are  kept in the query.
> ???

I tested this code works fine.

> 
> Is this a bug in the API, or am I doing  wrong?
> 
> 
> Also, for debugging purposes, how can I access the actual  query sent to the 
> sql server?
> 


If you still have issues
You can add debug printing in line 310 of session.cpp in function 
session::single
Before line:

        dbi_result res=dbi_conn_query(conn,escaped_query.c_str());

Add:

        std::cerr << "[" << escaped_query <<"]" << std::endl;

Also if exception is thrown
you can request query() parameter of dbixx_error.


Artyom